Skip to content

API Endpoints

All browser requests go through nginx, which routes by path prefix to one of the services below. This page lists every endpoint the editor relies on.

Routing summary

Path (as seen by the browser) Service
/api/predict nlp-api
/api/process_graph unwrap-api
/api/process_and_save wrap-up-api
/api/* (everything else) backend
/* web (Vue 3 SPA)

backend

Base: the Node/Express service. Reads Triply credentials from environment variables.

GET /

Health/welcome message.

GET /api/getSources

Lists available source documents via SPARQL.

Response

{ "message": "Sources retrieved!", "sources": { "head": {...}, "results": {...} } }

The sources body is a SPARQL JSON result with iri, title, date, and editor per source, ordered newest first.

POST /api/getSource

Exports one source graph as Turtle.

Request { "iri": "<source graph iri>" } Response { "message": "Source retreived!", "source": "<turtle string>" }

POST /api/getTasksFromTriply

Lists tasks via SPARQL.

Request { "endpoint": "TNO/editor/sparql" } Response { "message": "Tasks retrieved!", "tasks": { ...SPARQL JSON... } } with iri, title, date, editor per task.

POST /api/getTask

Exports a task graph and every graph it calc:involves (interpretation + sources) as a single TriG document.

Request { "iri": "<task iri>" } Response (JSON-encoded string) { "message": "Task retrieved!", "task": "<trig string>" }

POST /api/saveTaskAtTriply

Uploads a task to TriplyDB, skipping graphs that already exist online.

Request { "task": "<trig string>" } Response { "message": "Task saved!" } on success (HTTP 200).

nlp-api

POST /api/predict

Predicts the constituents of an act frame for Dutch text.

Request

{
  "text": "de verwerkingsverantwoordelijke verwerkt persoonsgegevens",
  "model": "bertje_2022_e4"
}

model is optional. It names an entry in the service's model registry — bertje_2022_e4 (the default) or legal-bert-dutch-english. An unknown key, or an absent one, falls back to the default rather than erroring.

Response

{
  "text": "...",
  "model": "bertje_2022_e4",
  "predicted_entities": [
    ["de", "None"],
    ["verwerkingsverantwoordelijke", "Actor"],
    ["verwerkt", "Action"],
    ["persoonsgegevens", "Object"]
  ]
}

Labels are Action, Actor, Object, Recipient, or None. The model field echoes the model actually used, which is not always the one requested — an unrecognised key returns the default's name, so a caller can tell the difference rather than assuming its request was honoured.

Errors

Status Body When
500 { "text": "...", "error": "Model does not exist on filesystem." } The requested model resolves to a path that is not present. Model files are mounted rather than baked into the image, so a registry entry can name a model the deployment does not have

A Swagger UI is available at /swagger.

unwrap-api

Converts FLINT RDF to editor JSON. The editor uses /process_graph (routed from /api/process_graph); the others support its standalone download page.

Method & path Purpose
GET / Renders the graph-picker page
GET /get_graph_names List graph names from the Triply dataset
GET /download_graph/<graph_iri> Download a graph as Turtle (gzip handled)
POST /process_graph Convert posted Turtle (Content-Type: text/turtle) into editor JSON
GET /process_and_download_graph/<graph_iri> Download a graph and return converted JSON

POST /process_graph — body is a Turtle string; response is the editor JSON interpretation.

wrap-up-api

Converts an editor JSON interpretation to FLINT RDF and stores it.

Method & path Purpose
POST /process_and_save (routed) Accepts an editor JSON interpretation, returns the FLINT RDF (Turtle/TriG)

The editor calls this via convertToRDF before downloading a .trig file or saving to TriplyDB.

Notes

  • The frontend sends an X-API-KEY header (from the X_API_KEY build variable) on the predict and process_and_save calls.
  • CORS is enabled on the backend and the Python services.
  • Error responses follow the usual HTTP status codes (401, 404, 500); the frontend surfaces them through an alert widget.