I haven't seen a package that binds to window in a long time. Better to make this a proper JS module, and let us do the binding ourself. It's 2021. We have great bundler tools now.

Hi there, thanks for putting this together! From my initial tests this thing is very fast. Loving it. I've been looking to integrate this with some of the work we're doing and ran into an issue. I'm hoping you can provide some advice.

The method getMatchingRecord has logic to determine if a relationship has been marked as a parent (already populated with data). If it has, rather than populating it's related fields again and returning that, it returns the relationship itself. I see why the logic is written this way. It prevents an infinite loop. Especially handy for potentially infinitely nested relationships.

The issue that I've run into is, I've got multiple examples where a relationship could be related to multiple things. So the first time it goes through this, it returns the populated relationship, but subsequent times it returns the plain relationship. For example:

Say you have a group that can have any number of users. And say you can have many groups. Users may belong to more than one group. Now let's say you have a user and this user has a relationship (like an avatar, or whatever else).

If this user is a member of multiple groups, the first group gets the user populated (and at that time marked as a parent, since they have relationships). But when we run through the second group the method will return the relationship when it runs into the same user again. Looking for any advice you can give on how to handle a situation like this. Thank you!

The optional links object on the top level of the response is missing from the parsed result. The links object should be added to the parsed result, but not modified in any way from the original version.

A pre-parsed object

{
  "links": {
    "next": "https://example.com?page=2"
  },
  "data": [
    { "id": "1", "type": "person", "attributes": { "name": "Person 1" } }
  ],
  "included": [],
  "meta": {}
}

Should look like this when parsed

{
  "links": {
    "next": "https://example.com?page=2"
  },
  "data": [
    { "id": "1", "type": "person", "name": "Person 1" }
  ],
  "included": [],
  "meta": {}
}

After parsing, I'm missing the type property on my entities. I'm wondering if all the non-attributes should be attached at something like meta so they're not completely lost, but don't collide with the attributes.

e.g.

{
  "data": [
    {
      "id": 1,
      "type": "person",
      "links": {},
      "attributes": {
        "name": "fred"
      }
    }
  ]
}

becomes

{
  "data": [
    {
      "id": 1,
      "name": "fred",
      "meta": {
        "type": "person",
        "links": {}
       }
    },
  ]
}

This is a very useful module but I have a JSONAPI response have data like this:

    "type": "articles",
    "id": "sf4379f1-bd3b-414d-8a17-191e17619d11",
    "attributes": {
      "id": "a_test_article_id"
    },

Then what I get will be:

    "type": "articles",
    "id": "sf4379f1-bd3b-414d-8a17-191e17619d11"

The attributes id is gone. That is caused by flatten function https://github.com/mysidewalk/jsonapi-parse/blob/master/src/jsonapi.js#L171

        return extend(
            {},
            { links: record.links, meta: meta },
            record.attributes,
            { id: record.id, type: record.type }
        );

That extend function merged all values by removing duplicated keys.
One question, do we really need to flatten the data object? What is the purpose?
In my case, there is a id field inside attributes but I can't get it after flatten.

Or if we have to flatten it, what about add attributes_id & attributes_type if there are such field inside attributes? Simply can be fixed by below:

    function flatten(record, extraMeta) {
        var meta = extend({}, record.meta, extraMeta)

       // Add this
        if (record.attributes.id) {
            record.attributes.attributes_id = record.attributes.id
        }

        if (record.attributes.type) {
            record.attributes.attributes_type = record.attributes.type
        }
       // End

        return extend(
            {},
            { links: record.links, meta: meta },
            record.attributes,
            { id: record.id, type: record.type }
        );
    }

So the attributes.id & attributes.type gets preserved.
Thanks.

Our data model has a middle table without any exposed attributes. e.g. order_item -> reservation -> serial
I need the serial number, but loading the serial entity as included entity, results in an error.

jsonapi.js (line 125)
Tries to add field on undefined:
record.attributes[property] = getMatchingRecord whereas attributes of record reservation is undefined. The only field ist the serial reference, which is not an attribute.

I have a relationship which returns some data in it. This data is not preserved and merged with the data from the resulting "includes".

"relationships": {
  "field_primary_image": {
    "data": {
      "type": "file--file",
      "id": "1234",
      "meta": {
        "alt": "My alt text",
        "title": "",
        "width": "1920",
        "height": "1080"
      }
    }
  }
}

The resulting field_primary_image contains the file/image data from the "includes" but does not have the meta data from the original relationships object with my alt text.