A CKAN extension for semantic data management in research data portals. It automatically pushes RDF resources to an Apache Jena Fuseki triple store, provides a CKAN-authenticated SPARQL proxy that enforces dataset-level access control, and embeds a YASGUI query interface for interactive SPARQL exploration of linked data.
- Apache Jena Fuseki server with custom security configuration (see Deployment section)
- CKAN API Token with admin privileges for background job processing of private datasets
This extension requires a custom Fuseki deployment with modified security settings to enable CKAN-authenticated access. The necessary files are provided in the optional/ folder:
optional/fuseki/Dockerfile- Custom Fuseki image based onsecoresearch/fuseki:6.2.0optional/fuseki/docker-entrypoint.sh- Entrypoint wrapper that starts Fuseki with the image's server classoptional/fuseki/shiro.ini- Shiro configuration that requires the Fusekiadminlogin for everything except the health endpoints; this is what blocks anonymous accessoptional/docker-compose.yml- Example standalone deployment configuration
See the optional/README.md for standalone deployment instructions, or integrate into your existing Docker Compose setup (see Deployment section below).
Optional: A Sparklis web app for interactive SPARQL querying is also available in the optional folder.
- Per-resource upload toggles — selectively push RDF resources (Turtle, N-Triples, JSON-LD, etc.) to Fuseki
- Named graphs — each resource is stored in its own named graph within a dedicated Fuseki dataset
- CKAN-authenticated SPARQL proxy — enforces CKAN dataset-level access control on all SPARQL endpoints
- Public datasets: accessible to anyone via the CKAN proxy
- Private datasets: require CKAN authentication (browser session or API token)
- Direct Fuseki access: blocked for anonymous users, admin-only
- Query Dataset button — embedded YASGUI interface for interactive SPARQL exploration
- Persistent & reasoning options — configurable at dataset level
See SECURITY.md for the full security architecture.
| CKAN version | Compatible? |
|---|---|
| 2.9 and earlier | not tested |
| 2.10 | ✓ tested |
| 2.11 | ✓ tested |
| 2.12 | ✓ tested |
From PyPI:
pip install ckanext-fusekiFrom Source:
pip install -e git+https://github.com/Mat-O-Lab/ckanext-fuseki.git#egg=ckanext-fusekiUse secoresearch/fuseki:6.2.0 (Jena 6, Java 21) with the provided configuration files mounted as volumes:
services:
fuseki:
# builds secoresearch/fuseki:6.2.0 plus the shiro.ini that blocks anonymous access
build:
context: ./path/to/ckanext-fuseki/optional/fuseki/
environment:
- JDK_JAVA_OPTIONS=-Xmx10g -Xms10g -DentityExpansionLimit=0 -Dfuseki:allowAddByConfigFile=true
- ADMIN_PASSWORD=<your-fuseki-password> # must match ckanext.fuseki.password
- ENABLE_DATA_WRITE=true
- ENABLE_UPDATE=true
- ENABLE_UPLOAD=true
- QUERY_TIMEOUT=60000
volumes:
- jena_data:/fuseki-base
networks:
- your_network
healthcheck:
test: ["CMD-SHELL", "wget -qO /dev/null http://localhost:3030/$$/ping || exit 1"]
interval: 10s
timeout: 5s
retries: 3shiro.ini— configures HTTP Basic Auth (the admin password is set fromADMIN_PASSWORDat start). Use the provided image build: the stocksecoresearch/fusekiimage ships ashiro.inithat allows anonymous SPARQL queries and copies it to the volume only on first start, so the stock image alone leaves all datasets public. The wrapper inoptional/fuseki/copies the restrictiveshiro.inito/fuseki-base/shiro.inion every start (the file must stay writable, the base entrypoint writes the password into it) and refuses to start without it. Verify withcurl http://fuseki:3030/ds/sparql?query=ASK%7B%7D(must return 401). The file must contain theDefaultWebSessionManagerlines fromoptional/fuseki/shiro.ini(Fuseki ≥ 5 requirement).- The Fuseki UI is served from the image itself and works behind an nginx
/fuseki/prefix without a customindex.html(Fuseki ≥ 5 uses relative asset paths).
Start Fuseki:
docker-compose up -d fusekiTested with secoresearch/fuseki:6.2.0 (Jena 6, Java 21): UI behind an nginx /fuseki/ prefix, admin login, dataset creation and a full update job. A TDB2 database created with 4.9 was opened unchanged by 5.6.0; an existing database was not re-tested on 6.2.0. Back up the /fuseki-base volume before upgrading. Required changes compared to Fuseki 4.9:
shiro.ini: add to[main]Without it every request that passes authentication returns HTTP 500statelessSessionManager = org.apache.shiro.web.session.mgt.DefaultWebSessionManager securityManager.sessionManager = $statelessSessionManager
No SessionManager.- Remove the
index.htmlmount (/jena-fuseki/webapp/index.html). The UI is served from inside the server jar and works behind an nginx/fuseki/prefix. - Dataset creation needs
-Dfuseki:allowAddByConfigFile=true. This extension creates one Fuseki dataset per CKAN dataset by posting an assembler description toPOST /$/datasets. Fuseki 5.5.0 removed that (multipart and raw body) because of CVE-2025-50151 / CVE-2025-49656 (uploaded configuration with unvalidated paths, admin only). The extension now posts the assembly as a rawtext/turtlebody, which Fuseki accepts only with this system property, e.g.JDK_JAVA_OPTIONS=-Xmx10g -Dfuseki:allowAddByConfigFile=true. Without it job creation fails with400 Service configuration from a multipart upload not supported. Keep Fuseki reachable only for admins (Shiro) and by CKAN; the Jena maintainers have said this switch is a stop-gap. Versions up to 5.4.0 accept the old upload without the property but are affected by the two CVEs. - The image default command is
org.apache.jena.fuseki.main.cmds.FusekiServerCmd. If you overrideentrypoint:in compose you must repeat the command, because Compose drops the imageCMD.
Datasets created with persistent off are in-memory and are empty after a Fuseki restart; resubmit them from CKAN.
Add fuseki to the ckan.plugins setting in your CKAN config file:
ckan.plugins = ... fusekidocker-compose restart ckan
# Or for Apache deployment:
sudo service apache2 reloadAll configuration is read from ckan.ini (or the equivalent CKANINI__ Docker environment variables):
| ckan.ini key | Default | Description |
|---|---|---|
ckanext.fuseki.url |
http://fuseki:3030/ |
Base URL of Fuseki used for all API calls from CKAN (dataset creation, uploads, queries, SPARQL proxy). The internal container address is recommended; a public URL behind nginx works too but adds a hop and needs the /fuseki/ route to forward the admin login |
ckanext.fuseki.internal_url |
(empty) | Optional direct container address, used only for the availability (ping) check; falls back to ckanext.fuseki.url |
ckanext.fuseki.username |
admin |
Fuseki admin username — always admin (hardcoded by the Fuseki image) |
ckanext.fuseki.password |
admin |
Fuseki admin password — must match the ADMIN_PASSWORD env var set on the Fuseki container |
ckanext.fuseki.ckan_token |
(empty) | CKAN API token for background job callbacks; required to process private datasets |
ckanext.fuseki.formats |
turtle text/turtle n3 nt hext trig longturtle xml json-ld ld+json jsonld |
Space-separated list of resource formats that trigger upload to the triple store |
ckanext.fuseki.ssl_verify |
true |
Verify SSL certificates when connecting to external resource URLs |
ckanext.fuseki.sparklis_url |
(empty) | If set, the query button redirects to this Sparklis instance instead of the built-in SPARQL UI |
ckanext.fuseki.db_url |
CKAN's sqlalchemy.url |
SQLAlchemy URL of the database that stores the job status and job log. Normally leave unset. Replaces the former FUSEKI_SQLALCHEMY_URL environment variable (which is now ignored). |
Example Docker environment variables (set in .env):
CKANINI__CKANEXT__FUSEKI__URL=http://fuseki:${FUSEKI_PORT}/
CKANINI__CKANEXT__FUSEKI__USERNAME=admin
CKANINI__CKANEXT__FUSEKI__PASSWORD=<your-fuseki-password>
CKANINI__CKANEXT__FUSEKI__CKAN_TOKEN=<your-ckan-api-token>
CKANINI__CKANEXT__FUSEKI__SSL_VERIFY=${SSL_VERIFY}Prefer the container-internal address (e.g. http://fuseki:3030/) for ckanext.fuseki.url: CKAN uses it for all direct Fuseki API calls and the SPARQL proxy, so a public nginx URL puts every call through the proxy.
If ckanext.fuseki.ckan_token is not set, only public resources can be uploaded to the triple store.
If you use this software, please cite it. GitHub shows a "Cite this repository" button (top right of the repo page) that exports the CITATION.cff in APA or BibTeX format.
After the first Zenodo release, a DOI-specific BibTeX entry will be available on the Zenodo record. Until then:
@software{hanke_ckanext_fuseki,
author = {Hanke, Thomas},
title = {ckanext-fuseki},
url = {https://github.com/Mat-O-Lab/ckanext-fuseki},
license = {AGPL-3.0-or-later},
}This project's work is based on a fork of the repo etri-odp/ckanext-jena, and we like to thank the authors of that project for sharing their work. It was supported by Institute for Information & communications Technology Promotion (IITP) grant funded by the Korea government (MSIT) (No.2017-00253, Development of an Advanced Open Data Distribution Platform based on International Standards)
