forest-express-mongoose
forest-express-mongoose copied to clipboard
ForestAdmin
[+] Forms - Embedded arrays containing "deep" embedded documents are json inputs
Expected behavior
Display array of embedded documents as related data
Actual behavior
Displayed as json in the parent document
Context
I have a Society with an array of Member embedded. Members are displayed in "details" and not in Related Data. I tried to create a fake field as a workaround with:
{ field: '_members', type: ['String'], reference: 'Society.members' }
but it try to display societies (whereas I guessed the model put in Liana.ResourceSerializer(Liana, Member... would say to forest "hey it's a Member")
- Package Version: 2.13.1
- Express Version: 4.15.4
- Mongoose Version: 4.10.8
- MongoDB Version: 3.4.9
Hey @Listu we were wondering if you could give us a sample of data you had in your database. A Society document with the embedded members. Thanks !
Of course @root-io Hope it helps !
` {
"_id" : ObjectId("5b8fde138e931929505a087a"),
"updatedAt" : ISODate("2018-09-17T09:11:03.086Z"),
"createdAt" : ISODate("2018-09-05T13:45:56.398Z"),
"name" : "SPT",
"address" : {
"place_id" : "ChIJfXRynsNv5kcRZ9ypL-zaaoI",
"country" : "France",
"city" : "Paris",
"area" : "Paris",
"street" : "37 Rue Marbeuf",
"postal_code" : "75008",
"coordinates" : [
48.86967500000001,
2.305317000000059
],
"_id" : ObjectId("5b8fde138e931929505a0882")
},
"invitations" : [ ],
"status" : 0,
"stripe" : "**************",
"billingFrequency" : 2,
"billings" : [
{
"_id" : ObjectId("5b8fde138e931929505a0878"),
"address" : {
"place_id" : "ChIJfXRynsNv5kcRZ9ypL-zaaoI",
"country" : "France",
"city" : "Paris",
"area" : "Paris",
"street" : "37 Rue Marbeuf",
"postal_code" : "75008",
"coordinates" : [
48.86967500000001,
2.305317000000059
],
"_id" : ObjectId("5b8fde138e931929505a0879")
},
"nic" : "63214",
"siren" : "123654789",
"capital" : 15000,
"raison" : "SPT",
"service" : null
},
{
"_id" : ObjectId("5b9f6ee870be0f42d0a0384e"),
"address" : {
"place_id" : "ChIJfXRynsNv5kcRZ9ypL-zaaoI",
"city" : "Paris",
"area" : "Paris",
"country" : "France",
"postal_code" : "75008",
"street" : "37 Rue Marbeuf",
"coordinates" : [
48.86967500000001,
2.305317000000059
],
"_id" : ObjectId("5b9f6ee870be0f42d0a0384f")
},
"capital" : 10000,
"nic" : "12355",
"siren" : "123456789",
"raison" : "TEST",
"service" : "Yolo"
}
],
"members" : [
{
"updatedAt" : ISODate("2018-09-05T13:45:56.396Z"),
"createdAt" : ISODate("2018-09-05T13:45:56.396Z"),
"user" : ObjectId("5b8fde128e931929505a085d"),
"billing" : ObjectId("5b8fde138e931929505a0878"),
"politic" : {
"_id" : ObjectId("5b8fdd468e931929505a0776"),
"rental" : {
"maxprice" : 100
},
"price" : {
"min" : 0,
"max" : 500,
"_id" : ObjectId("5b8fdd468e931929505a0777")
},
"stars" : {
"min" : 1,
"max" : 5,
"_id" : ObjectId("5b8fdd468e931929505a0778")
}
},
"_id" : ObjectId("5b8fde138e931929505a087b"),
"roles" : [
"executive"
]
},
{
"updatedAt" : ISODate("2018-09-17T09:11:03.085Z"),
"createdAt" : ISODate("2018-09-17T09:10:52.029Z"),
"user" : ObjectId("5b9f6f9b70be0f42d0a03881"),
"politic" : {
"flight" : {
"price" : {
"max" : 1100,
"min" : 0
},
"class" : "business"
},
"_id" : ObjectId("5b9f6fa770be0f42d0a03aac"),
"rental" : {
"maxprice" : 100
},
"price" : {
"min" : 0,
"max" : 500,
"_id" : ObjectId("5b9f6fa770be0f42d0a03aad")
},
"stars" : {
"min" : 2,
"max" : 5,
"_id" : ObjectId("5b9f6fa770be0f42d0a03aae")
}
},
"billing" : ObjectId("5b9f6ee870be0f42d0a0384e"),
"_id" : ObjectId("5b9f6f9c70be0f42d0a03888"),
"roles" : [
"traveler"
]
}
],
"__v" : 1
} `
Hi @Listu, this is currently a limitation of our platform.
In the case of arrays of embedded document that contain only keys/values with a "single depth level", these arrays will appear in the "Related Data" section.
For more complex documents in embedded arrays, as the edition may be very complex, we display a JSON field to manipulate the data. This is clearly not an ideal usage, but at least you can read and manipulate the data. We definitely need to work on this.
Ok so we have to change our model, wait an improvement or dev our own dashboard. Thank you for the help.
We have the same problem over here. Actually this can be solved with a smart collection of sub-documents, with a one-to-many relationship between the parent document and the sub-documents, by implementing the json-api. The "id" of each subdocument should contain both the the id of the parent document and the id of the sub-document to avoid collisions between two sub-documents with the same id that reside in two different parent documents. This can be skipped if the sub-documents' ids are unique (i.e. when they are autogenerated by mongoose). Also, each subdocument should have a reference to its exactly-one parent document, and an index for its position inside the parent document array. The references themselves the be implemented with Liana smart relationships. Paging can also be a non-trivial because of the ordering. It could be solved by sorting the subdocuments first by the parent-id and then by their own subdocument-id.
Flattening all the models could annoying, especially if you are actually modeling a tree structure, i.e. "Project" containing an array of "Sections", containing an array of "Tasks". This is one of the greatest advantages of NoSql - you can simply load the document with Robo3t or other clients, load the document, and see the tree-structure without needing to write code for populating all the arrays of references.
