Calling Manta from a backend

Everything the React SDK does goes through two HTTP APIs that you can call from any language: GraphQL for operations such as rules and action plans, and SPARQL for reading the knowledge graph.

Authenticate

Requests carry a Keycloak access token as a bearer token. A backend gets one of two ways, both through a custom app:

• Machine access. The app’s server mints its own token with the client ID and secret. It acts for the organization that owns the app, with a member’s access to its datastores.

  import { cachedToken, mintWithClientCredentials } from "@manta-eu/auth/web";
  
  const getAccessToken = cachedToken(() => mintWithClientCredentials({ clientId, clientSecret }));

• A person’s token. The server signs a person in and stores their refresh token, then calls Manta as them. See Backend for frontend.

GraphQL

Send operations to /graphql on your Manta deployment (by default https://app.hiremanta.com). Name the datastore in the x-manta-datastore header:

curl https://app.hiremanta.com/graphql \
  -H "Authorization: Bearer $MANTA_TOKEN" \
  -H "x-manta-datastore: your-datastore" \
  -H "Content-Type: application/json" \
  -d '{"query": "query { rules(limit: 5) { count rules { ruleIri committedRule } } }"}'

The rules and actions pages show the payloads for each operation.

SPARQL

Read the graph with a SPARQL query posted to /sparql with the x-manta-datastore header, or to the datastore’s own path, /d/<datastore>/sparql; both work the same. The endpoint accepts read queries only and answers anything else, such as INSERT DATA, with a 403. Write to the graph by connecting a source or uploading files, or through the GraphQL operations above.

curl https://app.hiremanta.com/d/your-datastore/sparql \
  -H "Authorization: Bearer $MANTA_TOKEN" \
  -H "Content-Type: application/sparql-query" \
  -H "Accept: application/sparql-results+json" \
  --data 'SELECT ?s WHERE { ?s a ?type } LIMIT 10'

Pages in this chapter

• Service account
• GraphQL
• SPARQL