Nanopub Query

Nanopub Query is the second-generation query service for nanopublications.

Available Instances

You can check out Nanopub Query at these instances:

• https://query.knowledgepixels.com/
• https://query.nanodash.net/
• https://query.petapico.org/

The current status of the live instances is shown at https://monitor.knowledgepixels.com/.

Documentation

See the JavaDocs for the API and source code documentation.

Repos and Triples

Each nanopublications is loaded into different repos in the form of RDF4J triple stores. There are these general repos:

• meta: Stores some specific "admin-graph metadata" of the nanopublications (see below), but not the nanopublications themselves
• full: Stores all nanopublications in full and their admin-graph metadata (this is not scalable on the long term, so will be deprecated in the medium-term future)
• last30d: Stores all nanopublications of the last 30 days and their admin-graph metadata
• text: Stores all nanopublications for full-text search
• admin: Stores some further admin info, such as the full pubkeys for their hash values
• empty: Empty repo from which other repos can be accessed via the SPARQL service keyword

On top of that, there are these specific repos:

• pubkey: For each public key used to sign a nanopublication, a separate repo is created
• type: For each nanopub type, a separate repo is created too

Two admin graphs (npa:graph and npa:networkGraph) are created with metadata about the nanopublications. The admin triple table shows the details.

Running an Instance

A Query instance loads its nanopublications from a Nanopub Registry, ideally one you run next to it. It runs with Docker Compose, as two containers: the Query app itself and the RDF4J triple store behind it.

Create the data directories, copy the docker-compose.override.yml.template file to docker-compose.override.yml, and adjust the settings in its Section 1 (public URL and registry):

./init-dirs.sh
cp docker-compose.override.yml.template docker-compose.override.yml

Then start it:

docker compose up -d

The Query app serves plain HTTP on localhost:9393 (plus a metrics endpoint on localhost:9394), and RDF4J on localhost:8081. None of these have authentication, so they are bound to localhost only. To make the instance publicly reachable via HTTPS, run a reverse proxy on the host that terminates TLS: nginx.conf is the reference configuration, which also restricts the SPARQL endpoints to read-only queries and answers CORS preflight requests.

On first start, the instance loads all nanopublications from the registry, which takes a while. Every response reports the progress in its headers: Nanopub-Query-Status goes from LOADING_INITIAL to READY, and Nanopub-Query-Loaded-Nanopub-Count approaches the registry's count in Nanopub-Query-Registry-Nanopub-Count:

curl -sI http://localhost:9393/ | grep -i '^nanopub-query'

Stop the instance with docker compose stop or docker compose down only, and give it time: the stack is configured to shut down its store cleanly, which can take a few minutes. Killing the containers mid-write can corrupt the store, and on a typical Ubuntu server two host settings need adjusting so that reboots and automatic updates do not do exactly that.

See the operations guide for hardware requirements, preparing the host, upgrading, monitoring, and repairing an instance.

Development

To build and run from the local sources, run:

./run.sh

Optional development features like remote JVM debugging (localhost:5005) can be enabled in the development section of docker-compose.override.yml.

License

This software is made available under the MIT license. See LICENSE.txt for the details.

For an overview of the dependencies and their licenses, run mvn project-info-reports:dependencies and then visit target/reports/dependencies.html.