pycottas is a library for working with compressed RDF files in the COTTAS format. COTTAS stores triples as a triple table in Apache Parquet. It is built on top of DuckDB and provides an HDT-like interface.
- Compression and decompression of RDF files.
- Querying COTTAS files with triple patterns.
- RDFLib store backend for querying COTTAS files with SPARQL.
- Supports RDF datasets (quads).
- Can be used as a library or via command line.
- Serve COTTAS files as an SPARQL endpoint.
PyPI is the fastest way to install pycottas:
pip install pycottasWe recommend to use virtual environments to install pycottas.
import pycottas
from rdflib import Graph, URIRef
pycottas.rdf2cottas('my_file.ttl', 'my_file.cottas', index='spo')
res = pycottas.search('my_file.cottas', '?s <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?o')
print(res)
pycottas.cottas2rdf('my_file.cottas', 'my_file.nt')
# COTTASDocument class for querying with triple patterns
cottas_doc = pycottas.COTTASDocument('my_file.cottas')
# the triple pattern can be a string (below) or a tuple of RDFLib terms
res = cottas_doc.search('?s <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> ?o')
# COTTASStore class for querying with SPARQL
graph = Graph(store=pycottas.COTTASStore('my_file.cottas'))
res = graph.query('''
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
SELECT DISTINCT ?s ?o WHERE {
?s rdf:type ?o .
} LIMIT 10''')
for row in res:
print(row)To execute via command line check the docs. It is also possible to serve COTTAS files as an SPARQL endpoint with pycottas-endpoint.
pycottas is available under the Apache License 2.0.
Universidad Politécnica de Madrid.
If you used pycottas in your work, please cite the ISWC paper:
@inproceedings{arenas2026cottas,
title = {{COTTAS: Columnar Triple Table Storage for Efficient and Compressed RDF Management}},
author = {Arenas-Guerrero, Julián and Ferrada, Sebastián},
booktitle = {Proceedings of the 24th International Semantic Web Conference},
year = {2026},
publisher = {Springer Nature Switzerland},
isbn = {978-3-032-09530-5},
pages = {313--331},
doi = {10.1007/978-3-032-09530-5_18},
}