datalab-app-xrayct¶

A datalab plugin for cataloguing X-ray computed tomography (CT) and operando battery imaging datasets from facilities like Diamond Light Source.

Why a plugin?¶

Tomography datasets routinely exceed 100 GB, so moving them to a web server is infeasible. This plugin uses a Control Plane / Data Plane split:

Control Plane (datalab): stores metadata, a small preview image, and a remote-storage URI.
Data Plane (HPC / deep storage): keeps the raw NeXus / TIFF volumes exactly where they are.

Phase 1 status (this release)¶

✅ NeXus / HDF5 metadata extraction (nexusformat + h5py)
✅ TIFF stack support (tifffile)
✅ Pydantic schema (XrayCTMetadata) with explicit URI scheme
✅ Central-slice PNG preview generation
✅ Diamond URI resolver (diamond://i13/<year>/<visit>/...)
✅ Mock NeXus generator for tests
✅ Vue 3 datacard component (webapp/XrayCTBlock.vue)

Install¶

pip install -e ".[local]"

The local extra pulls in datalab-server[apps] so the block registers via the pydatalab.apps.plugins entry point.

URI scheme¶

Lock in early — Phase 2 depends on it:

diamond://<beamline>/<year>/<visit>/<rest>
# e.g. diamond://i13/2025/mg39713-1/experiment/scan_00123.nxs
#  →  /dls/i13/data/2025/mg39713-1/experiment/scan_00123.nxs

Tests¶

pytest

Tests use a generated mock NeXus file (~few KB) and don't require any production data or a running datalab server.

Roadmap¶

See docs/ for the full Phase 1 → 4 plan (URI routing, Zarr/Dask interactive viewer, SLURM job dispatch).