wanted to ask if it is possible to upsert nested objects? for example, if i have a 'Users' table and a 'Students' table, and I'm inserting a new User(with a taken id), i want to update all fields (using on_conflict and update_columns) including the fields in the 'Students' table.
Basically replace all user's fields except the primary key.
mutation($UsersData: [core_users_insert_input!]!) {
insert_core_users(
objects: $UsersData
on_conflict: {
constraint: core_users_id_unique
update_columns: [first_name, last_name, gender]
}
) {
affected_rows
}
}
The update_column array should include fields from the 'Students' table but i can't figure it out.
It is possible, relevant documentation is here: https://hasura.io/docs/1.0/graphql/manual/mutations/upsert.html#upsert-in-nested-mutations
It is possible to use on_conflict key on any level (top, or nested) where you want to resolve updating an existing record.
mutation upsert_author_article {
insert_author(
objects: [
{
name: "John",
articles: {
data: [
{
title: "Article 3",
content: "Article 3 content"
}
],
on_conflict: {
constraint: article_title_key,
update_columns: [content]
}
}
}
]
) {
affected_rows
}
}
Related
I have two tables reels and reel_variations, reels can have many reel_variations and reel_variations belong to one reel. I have read the Hasura docs and haven't been able to figure out how to insert a reel and a couple of reel variations in a single mutation.
mutation insertReelsAndVariations($objects: [reels_insert_input!]! = {}) {
insert_reels(objects: $objects) {
affected_rows
returning {
description
id
name
variations {
ball_bearings
braid_capacity
created_at
deleted_at
gear_ratio
max_drag
line_capacity
id
model_number
recovery
reel_id
retrieve
}
}
}
}
Variables
{
"objects": {
"name": "nice reel",
"description": "wicked nice reel",
"variations": {
"data": {
"ball_bearings": "djjdfkjdkjfdjkfjkd",
"braid_capacity": "dkfjdkfjkdf",
"gear_ratio": "20:1",
"max_drag": "20lbs",
"line_capacity": "400yrds",
"model_number": "jfdkjfkjdkfjkdjfjdf",
"recovery": "30 per turn"
}
}
}
}
Errors
{
"errors": [
{
"extensions": {
"path": "$.selectionSet.insert_reels.args.objects[0].variations.data",
"code": "constraint-violation"
},
"message": "Not-NULL violation. null value in column \"reel_id\" violates not-null constraint"
}
]
}
That's because the reel_id column is not set as a foreign key col referencing the reels table. So u can simply:
Make the reel_id col a foreign key which points to the id column of reels table!
I am trying to do an upsert in a single mutation. Here I have two tables Users table [id,isVerified,type] and Customers table [id,name,deviceToken] Here,Customers.id is a foreign key of Users.Id.
The following is the mutation-
MyMutation {
insert_Users(objects: [{isVerified: false, name: "+9100000000", type: "customer",
Customers: {data: {deviceToken: "TestToken001"}}}],
on_conflict: {
constraint: Users_name_key,
update_columns: [isVerified]
}) {
affected_rows
returning {
Customers{
deviceToken
}
}
}
} ```
//But when I run this, I get the exception
{
"errors": [
{
"extensions": {
"path": "$.selectionSet.insert_Users.args.objects[0].Customers.data",
"code": "constraint-violation"
},
"message": "Uniqueness violation. duplicate key value violates unique constraint \"Customers_pkey\""
}
]
}
This seems to be because I am not setting conflict constraint on the nested Customers Object. How do I add the conflict constraint for a nested object?
You need to add the constraint object inside the nested data as well.
Something like:
MyMutation {
insert_Users(objects: [{isVerified: false, name: "+9100000000", type: "customer",
Customers: {
data: {deviceToken: "TestToken001"},
on_conflict: {
constraint: Customers_pkey,
update_columns: [deviceToken]
}
}}],
on_conflict: {
constraint: Users_name_key,
update_columns: [isVerified]
}) {
affected_rows
returning {
Customers{
deviceToken
}
}
}
}
I am building my first many-to-many insert mutation in Hasura and finding it difficult. The syntax in the docs and the accompanying explanation is very difficult to follow.
I am simply trying to add a connection between a component and a module.
Here is the state of my current query.
mutation MyMutation {
insert_component(objects: {component_module: {data: {module: {data: {id: "775c9e27-c974-4cfa-a01f-af50bd742726"}, on_conflict: {constraint: module_id_key, update_columns: id}}}}}) {
affected_rows
returning {
id
component_modules
}
}
}
Here is the error I get.
{
"errors": [
{
"extensions": {
"path": "$.selectionSet.insert_component.args.objects[0].component_module.data",
"code": "constraint-violation"
},
"message": "Not-NULL violation. null value in column \"component_id\" violates not-null constraint"
}
]
}
Here is my component table
Here is my module table
Here is my component_module bridge table
Thanks in advance for your help.
Your mutation is not working because you are inserting the id manually and when Hasura generates the query it won't have the id in the parent.
When doing nested inserts the best is to let PostgreSQL generate the ids for you. This way you will be able to insert with either side of the relationship.
In your example you don't really need to have the component_modules column in each table. When doing many to many inserts you can use the id of each table as the foreign key.
For example:
component
- id
- created_at
- updated_at
- name
module
- id
- created_at
- updated_at
- name
component_module
- component_id
- module_id
And the mutation should be something like:
mutation {
insert_component(objects: {
name:"component name",
component_modules: {
data: {
module: {
data: {
name: "module name"
}
}
}
}
}) {
returning {
id
component_modules {
component {
name
}
}
}
}
}
I have 2 tables users and post
Table users has columns id and post, column contains an array of the form [1, 2, 3, 4, 5] - where 1, 2, 3, 4, 5 is id in table post
In the table posts the following columns id and text
Table users:
https://i.stack.imgur.com/ywdS7.png
Table posts:
https://i.stack.imgur.com/IBdpb.png
in hasura made an array relation
https://i.stack.imgur.com/311sd.png
Next I made the following request
{
users_test {
postz {
id
}
}
}
I would like to receive such data in response:
postz: [
{
text: 'qwe'
},
{
text: 'sdf'
}
]
But with such a request, I get a trace. error:
{
"errors": [
{
"extensions": {
"internal": {
"statement": "SELECT coalesce(json_agg(\"root\" ), '[]' ) AS \"root\" FROM (SELECT row_to_json((SELECT \"_5_e\" FROM (SELECT \"_4_root.ar.root.postz\".\"postz\" AS \"postz\" ) AS \"_5_e\" ) ) AS \"root\" FROM (SELECT * FROM \"public\".\"users_test\" WHERE ('true') ) AS \"_0_root.base\" LEFT OUTER JOIN LATERAL (SELECT coalesce(json_agg(\"postz\" ), '[]' ) AS \"postz\" FROM (SELECT row_to_json((SELECT \"_2_e\" FROM (SELECT \"_1_root.ar.root.postz.base\".\"id\" AS \"id\" ) AS \"_2_e\" ) ) AS \"postz\" FROM (SELECT * FROM \"public\".\"posts\" WHERE ((\"_0_root.base\".\"post\") = (\"id\")) ) AS \"_1_root.ar.root.postz.base\" ) AS \"_3_root.ar.root.postz\" ) AS \"_4_root.ar.root.postz\" ON ('true') ) AS \"_6_root\" ",
"prepared": true,
"error": {
"exec_status": "FatalError",
"hint": "No operator matches the given name and argument type(s). You might need to add explicit type casts.",
"message": "operator does not exist: json = integer",
"status_code": "42883",
"description": null
},
"arguments": [
"(Oid 114,Just (\"{\\\"x-hasura-role\\\":\\\"admin\\\"}\",Binary))"
]
},
"path": "$",
"code": "unexpected"
},
"message": "postgres query error"
}
]
}
What am I doing wrong and how can I fix it?
A few suggestions:
There are some typos in your query, as far as I can tell. Try:
{
users {
id
posts {
text
}
}
}
You don't need the post column on the users table. You just need a user_id column on the posts table, and a foreign key constraint from the posts table to the users table using the user_id and id columns of the tables respectively. Check out the docs here:
https://docs.hasura.io/1.0/graphql/manual/schema/relationships/create.html#step-3-create-an-array-relationship
https://docs.hasura.io/1.0/graphql/manual/schema/relationships/database-modelling/one-to-many.html
If you have to have the post array column for some reason, you can use computed fields to create a "relationship" between a json array and another table’s id.
https://docs.hasura.io/1.0/graphql/manual/schema/computed-fields.html#table-computed-fields
Your function would:
Take in the json array column
Extract the id's
Return select * from table where id in id's
Example:
https://jsonb-relationships-hasura.herokuapp.com/console/api-explorer
Computed field definition at: https://jsonb-relationships-hasura.herokuapp.com/console/data/schema/public/tables/authors/modify
Run these queries:
# Get list of articles for each author
query {
authors {
id
name
articles
}
}
# Get actual articles for each author
query {
authors {
id
name
owned_articles {
id
title
}
}
}
Let's assume a type hierarchy of Customer -(hasMany)-> Orders -(hasMany)-> OrderLines
Something like this:
Customer {
Name
Orders [
{
OrderId
Date
OrderLines [
{
ItemCount
ItemName
}
]
}
]
}
I want to query for this whole tree, and filter on properties at any level in the tree.
For instance: Get all customers who ordered 'gizmos'.
This is what I tried: at each level of the hierarchy, I specify optional arguments that would filter based on the properties available at that level:
Customer (Name) {
Name
Orders (OrderId, Date) [
{
OrderId
Date
OrderLines (ItemCount, ItemName) [
{
ItemCount
ItemName
}
]
}
]
}
GraphQL needs me to define how to resolve each type in the hierarchy, so when resolving, I filter based on the arguments in the query.
But what if I only specify a filter at a deep level? e.g. ItemName : 'gizmo'
Assuming there's only one order line in the system containing a gizmo, I would expect to get a response like this:
[{
Name: "cust12",
Orders [{
OrderId: "ade32f",
OrderLines: [{
ItemCount: 50000, //customer really likes gizmos
ItemName: "gizmo"
}]
}]
}]
But what I actually get is all customers (no filter there), all their orders (no filter there) and all order items, mostly empty (the items inside are filtered).
[{
Name: "cust12",
Orders [
{
OrderId: "aaaaaa",
OrderLines: [ ]
},
{
OrderId: "ade32f",
OrderLines: [{
ItemCount: 50000,
ItemName: "gizmo"
}]
},
{
OrderId: "bbbbbb",
OrderLines: [ ]
},
{
OrderId: "cccccc",
OrderLines: [ ]
}
]
},
{
Name: "cust345",
Orders [
{
OrderId: "eeeeee",
OrderLines: [ ]
},
{
OrderId: "ffffff",
OrderLines: [ ]
}
]
}]
GraphQL calls the resolvers top-down:
- get all (filtered) clients
- for each of these get all (filtered) orders
- for each of those get all (filtered) order lines
Because of the top-down nature of calling the resolvers, I get a lot more data than I bargained for.
How should I approach this?
Relation filters
This is actually a more complex topic than it first seems. The problem is that your current filter condition expresses
get all customers, but only include items named 'gizmo'
but what you really want is
get all customers that are related to at least one item named 'gizmo'
get all customers that are related to at least one item named 'gizmo'
An elegant solution for this problem is the addition of relation filters to the schema. In your case, it could look like this:
query {
Customer(filter: {
orders_some: {
orderLines_some: {
item: {
itemName: "gizmo"
}
}
}
}) {
Name
Orders {
OrderId
Date
OrderLines {
ItemCount
ItemName
}
}
}
}
Using
orders_some: {
orderLines_some: {
item: {
itemName: "gizmo"
}
}
}
we only fetch customers that are indirectly related to an item named 'gizmo', exactly what we wanted.
Two more examples:
get all customers that are not related to any item named 'gizmo'
query {
Customer(filter: {
orders_none: {
orderLines_some: {
item: {
itemName: "gizmo"
}
}
}
}) {
Name
Orders {
OrderId
Date
OrderLines {
ItemCount
ItemName
}
}
}
}
get all customers where all their orders contain some order line with an item named 'gizmo'
query {
Customer(filter: {
orders_every: {
orderLines_some: {
item: {
itemName: "gizmo"
}
}
}
}) {
Name
Orders {
OrderId
Date
OrderLines {
ItemCount
ItemName
}
}
}
}
The every, some and none relation filters are an essential part of the Graphcool APIs - you can read more here.