Endpoints and plan for collections

[dlamoris]

• /orgs/{orgId}/collections

  • GET - return collections metadata
  • POST with slug - same as PUT on /orgs/{orgId}/collections/{collectionId} (slug is collectionId)

• /orgs/{orgId}/collections/{collectionId}

  • GET - return collection metadata
  • PUT - check referenced graphs for existence (can be branch, lock, scratch),
    • check user permission to update collection
    • create new collection with the referenced graphs, or replace existing collection if already exist, ex. request body

        <> dct:title "some_collection" .
        <> mms:collects <../repos/repo1/branches/b1>, <../repos/repo2/locks/lock1>, <../../org2/repos/repo3/scratches/s1> .
      

• /orgs/{orgId}/collections/{collectionId}/query

  • POST - sparql query - get the graphs for each collected branch/lock/scratch (error if doesn't exist)
    • add FROM <graph> and FROM NAMED <graph> into the query for each graph referenced

• /orgs/{orgId}/collections/{collectionId}/graph

  • GET - same as query, construct the graph of the union

    • CONSTRUCT { ?s ?p ?o . }
      FROM <graph1>
      FROM <graph2>
      ...
      WHERE {
        ?s ?p ?o .
      }
      

• permission checking options

  • option 1 - collection curator has responsibility
    • on collection put, check user has permission to update collection, also admin permission to collected graphs
    • on query, user just needs read permission to collection, permission to underlying graphs is redefined through the collection
  • option 2 - check underlying permission on each query
    • on collection put, user just needs permission to update collection, and read permission to collected graphs
    • on query, user must have read access to all graphs, otherwise error

[dlamoris]

JPL-Devin#1
JPL-Devin#9