TabularRDM

TabularRDM is a React Flow application for building a small tabular research data workflow. Users can load spreadsheet data, preview columns, describe those columns, create RDF metadata, inspect RDF in a store, and export an RO-Crate ZIP.

Running the App

Install dependencies:

npm install

Start the Vite development server:

npm run dev

Build the production bundle:

npm run build

Serve the built app with the local Node server:

npm run serve

The production server serves files from dist/ and proxies /qudt requests to https://qudt.org so QUDT vocabulary data can be fetched without browser CORS issues.

Main Files

package.json : Defines the Vite scripts and runtime dependencies. Important libraries include React, React Flow, xlsx, rdfstore, ro-crate, and jszip.

server.js : Minimal static file server for the production build. It also provides the /qudt proxy used by the QUDT service.

vite.config.js : Vite configuration for the React app.

index.html : Vite HTML entry point.

src/main.jsx : Mounts the React application.

src/App.jsx : Main workflow canvas. It defines node templates, connection behavior, data propagation between nodes, spreadsheet parsing, RDF propagation, and RO-Crate input propagation.

src/styles.css : Application-wide styles for the canvas, sidebar, and custom node UI.

Node Files

src/nodes/TabularFileNode.jsx : File input node. Reads a selected CSV or spreadsheet file as an ArrayBuffer and passes it to App.jsx.

src/nodes/PreviewTabularDataNode.jsx : Displays a preview table for the connected tabular file.

src/nodes/ColumnDescriptionNode.jsx : Lets users enter descriptions for detected columns. Units are assigned only by dragging a unit from the Units node. Changes are serialized to RDF automatically; there is no manual save button.

src/nodes/QuantityKindNode.jsx : Lets users search and select QUDT quantity kinds.

src/nodes/UnitNode.jsx : Lets users search QUDT units, filtered by a selected quantity kind when connected. Unit selections can be dragged into Column Description unit fields.

src/nodes/MetadataProfileSearchNode.jsx : Searches AIMS metadata profiles and passes selected profile information to a connected Metadata Form node.

src/nodes/MetadataFormNode.jsx : Wraps the SHACL form web component. It loads default metadata shapes or shapes from a selected AIMS profile and emits serialized RDF when the form is saved.

src/nodes/RDFStoreNode.jsx : Loads connected Turtle RDF into rdfstore-js, shows triple counts, previews the loaded Turtle, and allows downloading metadata.ttl.

src/nodes/ROCrateNode.jsx : Creates an RO-Crate ZIP using ro-crate and jszip. It includes data.json from connected RDF content, CSV exports for connected spreadsheet sheets under original_data/, and ro-crate-metadata.json.

Service Files

src/services/aimsApi.js : Fetches and normalizes AIMS application profile data.

src/services/columnDescriptionRdf.js : Serializes column description fields into Turtle RDF.

src/services/metadataShapesService.js : Loads metadata shape definitions, using a worker when available.

src/services/metadataShapes.worker.js : Worker implementation for metadata shape processing.

src/services/qudtService.js : Fetches and normalizes QUDT quantity kind and unit data.

RO-Crate Export Flow

1. Load a spreadsheet with a Tabular File node.
2. Connect it to downstream tabular nodes and to an RO-Crate node if sheet CSVs should be included.
3. Create RDF by connecting Metadata Form, Column Description, or RDF Store output into the RO-Crate node.
4. Press Download on the RO-Crate node.

If the loaded workbook contains a sheet named export_config, the RO-Crate node reads rows from that sheet to derive:

dataset_id
dataset_name
dataset_title
dataset_label
title
dataset_description
description
license

The exported ZIP name is based on dataset_id, dataset_name, or dataset. The crate title falls back through dataset_title, dataset_label, title, and the derived dataset id.

Notes

npm run build may show a Vite warning about chunks larger than 500 kB. This is expected because libraries such as rdfstore, xlsx, ro-crate, and jszip are large. The warning is not a build failure.