Use document data in Access Control function

twoxic3 years ago

16

const History: CollectionConfig = {
  slug: 'history',
  admin: {
    useAsTitle: 'id'
  },
  access: {
    read: isAdminOrHasAccessToResourceInHistory,
    create: () => true,
    update: () => true,
    delete: () => true
  },
  fields: [
    {
      name: 'event',
      type: 'relationship',
      relationTo: ['analyses', 'orthophotos', 'locations', 'species'],
      hasMany: false,
      required: true
    }
  ],
  timestamps: true
};

export default History;

export const isAdminOrHasAccessToResourceInHistory: Access<any, User> = ({
  req: { user }
}) => {
  if (!user) {
    return false;
  }
  if (user.roles.includes('admin')) return true;
  if (user.roles.includes('user') && user[collectionSlug]) {
    return {
      ['id']: {
        ['in']: user[collectionSlug]
      }
    };
  }
};

• 

  thedunco3 years ago

  The

  req

  object should also contain the

  payload

  object which you can use to query for any other documents!

• 

  twoxic3 years ago

  I don't think this will help me since I need to know the ID of the element first. According to the docs (

  https://payloadcms.com/docs/access-control/collections#read

  ) this is only available on findByID operations, and not on "find all" operations

  
  
  

  (

  id of document requested, if within findByID

  )

• 

  thedunco3 years ago

  I think you would just need to do this with access functions in each of the

  individual collections

  that are included in that relationship field (analyses, orthophotos, locations, species). The

  access: { read: ...}

  in the

  History

  collection will control if they can read the ENTIRE

  History

  collection, not the individual related ones. It seems to me like

  if you're trying to prevent them from viewing items in the history that they don't have access to

  you need to provide access functions on the individual collections for analyses, orthophotos, locations, and species.

  
  
  

  If I'm misunderstanding what you're trying to do I'll need a better explanation of the whole setup and what exactly you're trying to do. Hope this helps!

• 

  twoxic3 years ago

  You're right, that is how I'm doing it now and that does sort of work. However in that case the result of the

  /history?limit=2

  query would be this:

  
  

  {
    "docs": [
      {
        "id": "644b82d6fedb175d3f3347d9",
        "event": {
          "relationTo": "analyses",
          "value": {
            ...documentWhichUserHasAccessTo
          }
        },
        "createdAt": "2023-04-28T08:24:54.871Z",
        "updatedAt": "2023-04-28T08:24:54.871Z"
      },
      {
        "id": "644b8299fedb175d3f3347b8",
        "event": {
          "relationTo": "analyses",
          "value": "644b8292fedb175d3f334713"
        },
        "createdAt": "2023-04-28T08:23:53.049Z",
        "updatedAt": "2023-04-28T08:23:53.049Z"
      }
    ],
    "totalDocs": 55,
    "limit": 2,
    "totalPages": 14,
    "page": 1,
    "pagingCounter": 1,
    "hasPrevPage": false,
    "hasNextPage": true,
    "prevPage": null,
    "nextPage": 2
  }

  
  
  

  However, I want to show the latest 2 relevant (cases in which the user actually has access to the "event"

  history

  items.

  
  
  

  So ideally I'd make an

  access

  function on the

  history

  collection.

• 

  jarrod_not_jared

  3 years ago

  @399871533276987392

  I believe you could use an afterRead hook to filter out the docs that the user does not have access to. If the user does not have access to the doc it should only return the string id, else it should return the entire doc. Can you give that a shot and see if that does what you need?

• 

  twoxic3 years ago

  I've tried that and that does work partially. The only problem I see is I don't want to fetch the entire history (collection) every time.

  
  
  

  Say I just want the 4 latest items in the collection that the user has access to. I'd use

  ?limit=4

  which could give me no records (if I don't have access to the latest 4 history-items.).

  
  
  

  While I'm sort of

  rubber-duckying

  myself I get the thought of stripping the

  limit

  query in a

  beforeRead

  hook and manually slicing the filtered docs in an

  afterRead

  . I'll try that later!

• 

  jarrod_not_jared

  3 years ago

  I am trying to align myself with your issue.

  
  
  

  - You have a history collection, with access control.

  
  

  - You query the history collection with limit=4

  
  

  - You do not get docs back if your user does not have access to the first 4 docs?

• 

  twoxic3 years ago

  So I have a

  history

  collection which stores documents

  ("events")

  which can be of multiple types(e.g.

  analyses

  ,

  orthophoto

  ). Users have access to specific

  analyses

  and

  orthophotos

  , but not to all of them. The access control for this is setup and working.

  
  
  

  The userstory I have for this usecase is

  As a user I want to retrieve the 2 (or insert any other number) latest documents in the history collection which references a document which the user has access to.
  
  

  At the moment, querying history like

  /history?limit=2

  gives back:

  
  

  "docs": [
      {
        "id": "644b82d6fedb175d3f3347d9",
        "event": {
          "relationTo": "analyses",
          "value": {
            ...documentWhichUserHasAccessTo
          }
        },
        "createdAt": "2023-04-28T08:24:54.871Z",
        "updatedAt": "2023-04-28T08:24:54.871Z"
      },
      {
        "id": "644b8299fedb175d3f3347b8",
        "event": {
          "relationTo": "analyses",
          "value": "644b8292fedb175d3f334713"
        },
        "createdAt": "2023-04-28T08:23:53.049Z",
        "updatedAt": "2023-04-28T08:23:53.049Z"
      }
    ],

  
  

  In which

  docs[0]

  the user has access to and

  docs[1]

  the user doesn't. So that doesn't give me back

  2 (or insert any other number) latest documents in the history collection which references a document which the user has access to.
  
  

  To solve this I would need to write an

  access

  hook on the

  history

  collection itself. But I'd need to access the individual document in this function to check of which type (analyses, orthophoto) it is before I can check if the user has access to it.

  
  
  

  One step closer! I've removed the

  limit

  query on

  beforeOperation

  like:

  
  

  export const removeLimitFromArgs: CollectionBeforeOperationHook = async ({
    args,
    operation
  }) => {
    if (operation !== 'read') {
      return args;
    }
  
    if (args.limit) {
      args._limit = args.limit;
      args.limit = 999999; // Remove limit for now.
    }
  
    return args;
  };

  
  

  This way payload grabs all

  history

  docs. Next step is filtering the docs and apply the limit. One problem: There is no hook in which you can access the list of retrieved documents, like:

  afterOperation

  . Why isn't there one? How to fix it without this hook?

  
  
  

  So the hooks would become like this:

  
  

  hooks: {
      beforeOperation: [removeLimitFromArgs],
      afterOperation: [filterUnaccessibleEvents, addPreviouslyRemovedLimit]
    },

• 

  jarrod_not_jared

  3 years ago

  I think you are looking for the beforeRead and afterRead hooks:

  https://payloadcms.com/docs/hooks/collections#afterread
  
  

  also, you can set the limit to 0 and that will retrieve all docs. (instead of 99999)

• 

  twoxic3 years ago

  afterRead seems to fire for every document which is read, I've added this function to the

  afterRead

  hooks.

  
  

  export const filterUnaccessibleEvent: CollectionAfterReadHook = async ({
    doc // full document data
  }) => {
    console.log('afterRead');
    return doc;
  };

  
  
  

  And one function to

  beforeOperation

  which does

  console.log('beforeOperation')

  . This results in the following output:

  
  

  beforeOperation
  afterRead
  afterRead
  afterRead
  afterRead
  afterRead
  afterRead
  afterRead
  afterRead
  afterRead
  afterRead
  afterRead
  [many more here]

• 

  jarrod_not_jared

  3 years ago

  you are logging these by hitting the API? or by loading the admin panel?

• 

  twoxic3 years ago

  Hitting the API

  
  
  

  I've looked in the source and it is clear that afterRead runs for every document:

  https://github.com/payloadcms/payload/blob/master/src/collections/operations/find.ts
  
  

    result = {
      ...result,
      docs: await Promise.all(result.docs.map(async (doc) => {
        let docRef = doc;
  
        await collectionConfig.hooks.afterRead.reduce(async (priorHook, hook) => {
          await priorHook;
  
          docRef = await hook({ req, query, doc: docRef, findMany: true }) || doc;
        }, Promise.resolve());
  
        return docRef;
      })),
    };

• 

  jarrod_not_jared

  3 years ago

  yep yep

• 

  twoxic3 years ago

  I could see usecases for a

  afterOperation

  (this one!). Do you think that would be a welcome addition? I could make a PR.

• 

  jarrod_not_jared

  3 years ago

  I am following

  
  
  

  I think what you are saying makes sense

  
  
  

  and sounds like a good feature

  
  
  

  I am going to link to this convo in the

  #1102950643259424828

  channel

• 

  twoxic3 years ago

  Cool. Let me know if I can do some work on it.