Squashage

Squashes JSON into graph sausage. Feed it structured JSON records. It classifies each one, reconstitutes the lot into a deterministic RDF graph, and squashes the result into a single file you can actually serve; Turtle, TriG, N-Triples, N-Quads, or JSON-LD.

Get startedWalk-throughLive demoGitHub

You feed it JSON. It hands you back a graph. The graph has edges where you'd expect edges and types where you'd expect types, and it ends up in a single file you can stick on a webserver.

• Same JSON in, same graph out. No Math.random, no Date.now, no network after startup. Byte-identical across runs and machines.
• One file, no fan-out. A single build produces one serialized RDF file. Auto-derived instance/graph/vocabulary IRIs from _source.url. Auto-built JSON-LD @context from the produced quad set.
• Open the demo offline. The squashage-dag viz CLI emits a self-contained HTML document with a cosmos.gl WebGL graph browser inlined. Open it in any browser; no network, no node_modules required.

Quick install

git clone https://github.com/Studnicky/Squashage.git
cd Squashage && npm install && npm run build
npx squashage-dag build --config squashage.config.json

Where to look next

• Getting started — install and run a build
• Walk-through — one record's full journey through the DAG
• DAG — the run-scope + per-record DAGs in full
• Classifier cascade — the ten classifiers + the conflict resolver
• Architecture — module map + class lineage
• Live demo — interactive graph of the AONPRD fixture