aggregate with pointer relation match ?

[xedixermawan]

Hi,
is it possible to do aggregate query using match with pointer relation ?
I'm using latest branch version.
for instance request like this:

curl -X GET \
  -H "X-Parse-Application-Id: myAppId" \
  -H "X-Parse-Master-Key: myAppKey" \
  -H "X-Parse-REST-API-Key: ${REST_API_KEY}" \
  -G \
  --data-urlencode 'match={"user":{"$eq":{"__type":"Pointer","className":"_User","objectId":"NE962BXR9x"}}}' \
	--data-urlencode 'group={"objectId":null,"total":{"$sum":"$value"}}' \
  http://localhost:1337/parse/aggregate/Distance

the Distance collection have a field calls 'user' ( a pointer type which point to collection _User.)

the aggregate above resulting empty data {"results":[]}, meanwhile it should not

[dplewis]

A quick workaround is appending _p_ to the field name (Mongo)

const pipeline = [
    { group: { objectId: '$_p_pointerFieldName' } }
];

Edit: Postgres Supports { group: { objectId: '$pointerFieldName' } } works out of the box

[dplewis]

Closed via #4493

[xedixermawan]

Hi @dplewis , thanks for quick reply & implementation.
I tried your code change and it's work by appending _p_.
the thing I notice, the form of aggregate query is not consistent with the form of relational query ( a query using keyword where ).
for my query above , here is the change I made to make it work.

match={"_p_user":"_User$m7WeWjAQ3P"}
group={"objectId":"$_p_user","total":{"$sum":"$value"}}

[dplewis]

@xedixermawan In mongo especially there are multiple stages (20+) for aggregates so adding transforms for all of them would take some time. I guess we can transform the main 3 group, match, project. You want to open a PR to add pointer transform to match? I'll review it