| name | build-data-bundle |
| description | Use when producing the shippable skosdex dataset — merging openly-licensed canonical N-Quads into dist/bundle.nq.gz, generating dist/solr-docs.json, building the docs/ site, or understanding which schemes are included/excluded and why. |
| license | Apache-2.0 |
Build the data bundle
The bundle is the deployable dataset that the SPARQL + Solr container ships.
node scripts/skosdex bundle
node scripts/skosdex solr-docs
node scripts/skosdex demo
node scripts/skosdex site
node scripts/skosdex build
Prereq: the committed canonical.nq.gz files are Git LFS objects — run
git lfs pull after a fresh clone or bundle reads empty pointer files.
What bundle does
- Includes a scheme only if its
licenseClass is open/public-domain and
skosdex:bundle is not false. Everything else is excluded with a reason.
- Reads each scheme's gzipped
canonical.nq.gz, places its triples in a
named graph equal to its conceptSchemeURI, and prefixes blank node
labels per-scheme, so multiple schemes merge without collisions.
- Writes gzipped
dist/bundle.nq.gz plus dist/manifest.json listing
included/excluded schemes, licenses, and quad counts — check the manifest to
confirm no copyleft/NC data slipped in.
Outputs
| File | Committed? | Consumed by |
|---|
dist/bundle.nq.gz | no (gitignored) | Oxigraph (gunzipped + baked into the skosdex/sparql image) |
dist/solr-docs.json | no | Solr (posted by the solr-init container) |
dist/manifest.json | no | provenance / license audit |
dist/demo.html | no | standalone browser; copied to docs/index.html |
docs/ (index.html, bundle.nq.gz, solr-docs.json, manifest.json) | yes | GitHub Pages site |
Note: everything in dist/ is a gitignored build output. docs/ IS committed
(it's the published site) and holds plain (non-LFS) small files, because
GitHub Pages serves LFS pointers, not content.
Run bundle/build before docker compose build (see the run-endpoints
skill), and site to refresh the Pages demo.
