Distinct with custom field

Help
, ,
1

Consider the following data:

[
  [
    "2021-06-01T14:17:45.032122Z",
    "51342145",
    Ref(Collection("chatMessages"), "1241454")
  ],
  [
    "2021-06-01T14:17:40.338061Z",
    "51342145",
    Ref(Collection("chatMessages"), "245634545")
  ]
]

And the following indexes:

CreateIndex({
  name: 'chatMessagesBySenderId',
  source: Collection('chatMessages'),
  terms: [
    { field: ['data', 'senderId'] }
  ],
  values: [
    { field: ['data', 'createdAt'], reverse: true },
    { field: ['data', 'recipientId'] },
    { field: ['ref'] }
  ]
})

CreateIndex({
  name: 'chatMessagesByRecipientId',
  source: Collection('chatMessages'),
  terms: [
    { field: ['data', 'recipientId'] }
  ],
  values: [
    { field: ['data', 'createdAt'], reverse: true },
    { field: ['data', 'senderId'] },
    { field: ['ref'] }
  ]
})

I wanted to do:

Paginate(
  Distinct(
    Union(
      Match(Index("chatMessagesByRecipientId"), authUser.id),
      Match(Index("chatMessagesBySenderId"), authUser.id)
    )
  )
);

and I wanted it to return:

{
  data: [
    [
      "2021-06-01T14:17:45.032122Z",
      "51342145",
      Ref(Collection("chatMessages"), "1241454")
    ]
  ]
}

In this case, I wanted to use the data[1] which is the 51342145 for uniqueness.

I’m unsure if this can be achieved in another way, but the best solution I can think of is to accept a second argument for Distinct which is a callback that will return data that will be used for uniqueness:

Paginate(
  Distinct(
    Union(
      Match(Index("chatMessagesByRecipientId"), authUser.id),
      Match(Index("chatMessagesBySenderId"), authUser.id)
    ),
    Lambda(
      ['createdAt', 'id', 'ref'],
      Var('id')
    )
  )
);

In the case above, Distinct will call the callback which will return the Var('id') which it will use to check for uniqueness.

2

You could probably replace Distinct with Reduce and construct an object that recorded the first distinct id discovered.

3

I have tried it that way but I could not make it work.

4

This appears to do what you want:

Reduce(
  Lambda(
    ["acc", "value"],
    Merge(
      Var("acc"),
      ToObject(
        [[
          ToString(Select(1, Var("value"))),
          Var("value")
        ]]
      )
    )
  ),
  {},
  Paginate(
    Union(
      Match(Index("chatMessagesByRecipientId"), 123),
      Match(Index("chatMessagesBySenderId"), 123)
    )
  )
)

I used a static id based on the documents I created to test the query: replace 123 with authUser.id.

The important bit in the reducer is calling Merge to compose a new accumulator object for each iteration that includes each individual page item. Composing dynamic objects can’t be done with { Var("field"): Var("value") }, which is the approach I tried first, but ToObject makes it work.