I'm playing around with GraphQL, and I've run across the concept of connections and edges.
From what I understand it's not uncommon to see metadata on the connection, like the totalCount property in the following snippet.
type UserFriendsConnection {
pageInfo: PageInfo!
edges: [UserFriendsEdge]
totalCount: Int
}
My questions is whether it's OK to put arbitrary metadata on the edge also, and if the following would be a decent way to do that.
I felt like a query and a response would best illustrate what I'm looking for. It's the role property I want to place somewhere that makes sense.
I feel like it doesn't belong in the User type since the role describes the type of connection/relationship the User has with a Group.
# Query
{
me {
id
name
groupsConnection {
edges {
node {
id
name
membersConnection {
edges {
node {
id
name
}
role <--- HERE
}
}
}
role <--- HERE
}
}
}
}
# Response
{
"data": {
"me": {
"id": "1Edj3hZFg",
"name": "John Doe",
"groupsConnection": {
"edges": [
{
"node": {
"id": "bpQgdZweQE",
"name": "Fishing Team",
"membersConnection": {
"edges": [
{
"node": {
"id": "1Edj3hZFg",
"name": "John Doe"
},
"role": "ADMINISTRATOR" <--- HERE
},
{
"node": {
"id": "7dj37dH2d",
"name": "Rebecca Anderson"
},
"role": "MEMBER" <--- HERE
}
]
}
},
"role": "ADMINISTRATOR" <--- HERE
}
]
}
}
}
}
Connections are part of the Relay specification. Relay itself is a GraphQL client, although you can have a Relay-compliant GraphQL server without actually using Relay on the front end. According to the spec:
Edge types must have fields named node and cursor. They may have additional fields related to the edge, as the schema designer sees fit.
It's common enough to see additional fields on these types and it certainly makes sense. One word of caution though. If we have a User type, we may create a UserConnection and a UserEdge:
type UserConnection {
pageInfo: PageInfo!
egdes: [UserEdge!]!
}
type UserEdge {
cursor: String!
edge: User!
}
We can then use that connection type in all sorts of places...
type Query {
allUsers: UserConnection!
# other fields
}
type Group {
members: UserConnection!
# other fields
}
type User {
coworkers: UserConnection!
# other fields
}
However, if you add a field like role to UserEdge, that field will only make sense in the context of the members field on the Group type. It would have to return null or some dummy value in all other contexts, which can introduce unnecessary confusion.
That means, if you're going to introduce extra fields on your edge type that are relationship-dependent, you should probably create connection and edge types that are specific to that relationship:
type GroupUserConnection {
pageInfo: PageInfo!
egdes: [GroupUserEdge!]!
}
type GroupUserEdge {
cursor: String!
edge: User!
role: Role!
}
This way, you can still use a regular UserConnection for other fields and avoid clients unnecessarily requesting a role where there isn't one.
Related
With a schema like
schema {
query: QueryRoot
}
scalar MyBigUint
type Order {
id: Int!
data: OrderCommons!
kind: OrderType!
}
type OrderBook {
bids(limit: Int): [Order!]!
asks(limit: Int): [Order!]!
}
type OrderCommons {
quantity: Int!
price: MyBigUint! // where it doesn't matter whether it's MyBigUint or a simple Int - the issue occurs anyways
}
enum OrderType {
BUY
SELL
}
type QueryRoot {
orderbook: OrderBook!
}
And a query query { orderbook { bids { data { price } }, asks { data { price } } } }
In a graphql playground of my graphql API (and on the network level of my Apollo app too) I receive a result like
{
"data": {
"orderbook": {
"bids": [
{
"data": {
"price": "127"
}
},
{
"data": {
"price": "74"
}
},
...
],
"asks": [
{
"data": {
"price": "181"
}
},
{
"data": {
"price": "187"
}
},
...
]
}
}
}
where, for the purpose of this question, the bids are ordered in descending order by price like ["127", "74", "73", "72"], etc, and asks are ordered in ascending order, accordingly.
However, in Apollo, after a query is done, I notice that one of the arrays gets seemingly random data.
For the purpose of the question, useQuery react hook is used, but the same happens when I query imperatively from a freshly initialized ApolloClient.
const { data, subscribeToMore, ...rest } = useQuery<OrderbookResponse>(GET_ORDERBOOK_QUERY);
console.log(data?.orderbook?.bids?.map(r => r.data.price));
console.log(data?.orderbook?.asks?.map(r => r.data.price));
Here, corrupted data of Bids gets printed i.e. ['304', '306', '298', '309', '277', '153', '117', '108', '87', '76'] (notice the order being wrong, at the least), whereas Asks data looks just fine. Inspecting the network, I find that Bids are not only properly ordered there, but also have different (correct, from DB) values!
Therefore, it seems something's getting corrupted on the way while Apollo delivers the data.
What could be the issue here I wonder, and where to start debugging such kind of an issue? There seem to be no warnings from Apollo either, it seems to just silently corrupt the data.
I'm clearly doing something wrong, but what?
The issue seems to stem from how Apollo caches data.
My Bids and Asks could have the same numeric IDs but share the same Order graphql type. Apollo rightfully assumes a Bid and an Ask with the same ID are the same things and the resulting data gets wrecked as a consequence.
An easy fix is to show Apollo that there's a complex key to the Order type on cache initialization:
cache: new InMemoryCache({
typePolicies: {
Order: {
keyFields: ['id', 'kind'],
}
}
})
This way it'll understand that the Order entities Ask and Bid with the same ID are different pieces of data indeed.
Note that the field kind should be also added to the query strings accordingly.
Let's imagine we have GraphQL API that can return an object Entity with Id and Name properties and I requested Name only:
query {
entities {
name
}
}
And it returns
{
"data": {
"entities": [
{
"name": "Name1"
},
{
"name": "Name2"
}
]
}
}
But what if I want to have only the name of entities as a scalar type? In other words, I want to have something like:
{
"data": {
"entities": [
"Name1",
"Name2"
]
}
}
Is it possible to have such result without changes on the GraphQL API side? Aliases, Fragments, etc. GraphQL has a lot of built-in query capabilities, but none of the known me can return complex objects as scalar type.
what you're asking for is almost impossible if you don't want to change the type definition for Entities.
This: 👇🏽
Entity: id: int! name: String
entities(): [Entity]
returns an array of objects with keys name and id.
To achieve what you're asking you either change Entity to be just a string or have your client reduce that object to an array of just Entity names when they receive it.
They could do something like this:
const data = {
entities: [
{
name: 'Name1',
},
{
name: 'Name2',
},
],
};
const entityNames = data.entities.reduce(
(acc, curr) => [...acc, curr.name],
[]
);
console.log(entityNames);
I'm trying to wrap my head around GraphQL.
Right now I'm just playing with the public API of Artsy (an art website, playground at https://metaphysics-production.artsy.net). What I want to achieve is following:
I want to get all node types entities without declaring them by hand (is there a shortcut for this)?
I want every node with a field type from which I can read the type, without parsing through imageUrl etc. to fint that out.
What I constructed as of right now is this:
{
search(query: "Berlin", first: 100, page: 1, entities: [ARTIST, ARTWORK, ARTICLE]) {
edges {
node {
displayLabel
imageUrl
href
}
}
}}
Very primitive I guess. Can you guys help me?
TL;DR:
1) There is no shortcut, it's not something GraphQL offers out of the box. Nor is it something I was able to find via their Schema.
2) Their returned node of type Searchable does not contain a property for type that you're looking for. But you can access it via the ... on SearchableItem (union) syntax.
Explanation:
For question 1):
Looking at their schema, you can see that their search query has the following type details:
search(
query: String!
entities: [SearchEntity]
mode: SearchMode
aggregations: [SearchAggregation]
page: Int
after: String
first: Int
before: String
last: Int
): SearchableConnection
The query accepts an entities property of type SearchEntity which looks like this:
enum SearchEntity {
ARTIST
ARTWORK
ARTICLE
CITY
COLLECTION
FAIR
FEATURE
GALLERY
GENE
INSTITUTION
PROFILE
SALE
SHOW
TAG
}
Depending on what your usecase is, if you're constructing this query via some code, then you can find out which SearchEntity values they have:
{
__type(name: "SearchEntity") {
name
enumValues {
name
}
}
}
Which returns:
{
"data": {
"__type": {
"name": "SearchEntity",
"enumValues": [
{
"name": "ARTIST"
},
{
"name": "ARTWORK"
},
...
}
}
}
then store them in an array, omit the quotation marks from the enum and pass the array back to the original query directly as an argument.
Something along the lines of this:
query search($entities: [SearchEntity]) {
search(query: "Berlin", first: 100, page: 1, entities: $entities) {
edges {
node {
displayLabel
imageUrl
href
}
}
}
}
and in your query variables section, you just need to add:
{
"entities": [ARTIST, ARTWORK, ...]
}
As for question 2)
The query itself returns a SearchableConnection object.
type SearchableConnection {
pageInfo: PageInfo!
edges: [SearchableEdge]
pageCursors: PageCursors
totalCount: Int
aggregations: [SearchAggregationResults]
}
Digging deeper, we can see that they have edges, of type SearchableEdge - which is what you're querying.
type SearchableEdge {
node: Searchable
cursor: String!
}
and finally, node of type Searchable which contains the data you're trying to access.
Now, the type Searchable doesn't contain type:
type Searchable {
displayLabel: String
imageUrl: String
href: String
}
But, if you look at where that Searchable type is implemented, you can see SearchableItem - which contains the property of displayType - which doesn't actually exist in Searchable.
You can access the property of SearchableItem and get the displayType, like so:
{
search(query: "Berlin", first: 100, page: 1, entities: [ARTIST, ARTWORK, ARTICLE]) {
edges {
node {
displayLabel
imageUrl
href
... on SearchableItem {
displayType
}
}
}
}
}
and your result will look like this:
{
"data": {
"search": {
"edges": [
{
"node": {
"displayLabel": "Boris Berlin",
"imageUrl": "https://d32dm0rphc51dk.cloudfront.net/CRxSPNyhHKDIonwLKIVmIA/square.jpg",
"href": "/artist/boris-berlin",
"displayType": "Artist"
}
},
...
Im new to AWS AppSync however its been pretty easy to learn and understand.
Im trying to create a resolver that when the user runs getChore(id: "") it will return all the chore information. Which its successfully doing, the problem is within the chore there are two fields: createdBy & assignedTo which are linked to a user type.
type Chore {
id: ID!
title: String
desc: String
status: String
reward: Float
retryDeduction: Float
required: Boolean
createdDate: AWSDateTime
date: AWSDateTime
interval: String
assignedTo: User
createdBy: User
}
type User {
id: ID!
age: Int
f_name: String
l_name: String
type: Int
admin: Boolean
family: Family
}
within aws appsync in trying to attach a resolver to assignedTo: User and createdBy: User so my query will look like:
query getChore {
getChore(id: "36d597c8-2c7e-4f63-93ee-38e5aa8f1d5b") {
id
...
...
assignedTo {
id
f_name
l_name
}
createdBy {
id
f_name
l_name
}
}
}
however when i fire off this query im getting an error:
The provided key element does not match the schema (Service: AmazonDynamoDBv2; Status Code: 400; Error Code: ValidationException;
which i have researched and cant seem to find the correct soltuion.
The resolver im using is:
{
"version": "2017-02-28",
"operation": "GetItem",
"key": {
"id": $util.dynamodb.toDynamoDBJson($ctx.args.id),
}
}
return:
$util.toJson($ctx.result)
When you get the The provided key element does not match the schema error, it's because your request mapping template key doesn't match the primary key in DynamoDB. You can enable CloudWatch Logs in your Application settings to see exactly what was sent to DynamoDB.
I'm not able to know what's wrong with your template because your sample lacks some information, if you can answers the questions pertaining to your application:
- Where are the users stored? Are they stored in their own DDB table separate from the chores, and is the hash key on the users table id as well?
- In the chores table how do you know which user your chore is assignedTo or createdBy? Is there a user id stored on the chore DDB item?
- Is the request mapping template you posted corresponding to the resolver attached to Chore.assignedTo? If yes, using $ctx.args.id will actually do a GetItem based on the chore id not the user it's assigned to.
Finally, I reproduced your application and I was able to make it work with a few changes.
Prerequisites:
I have a chores and a users DynamoDB table with both having id as hash key. These two tables are mapped as datasources in AppSync.
I have one chore in the chores tables that looks like
{
"assignedTo": "1",
"createdBy": "2",
"id": "36d597c8-2c7e-4f63-93ee-38e5aa8f1d5b",
"title": "Chore1"
}
and two users in the users table:
{
"f_name": "Alice",
"id": "2",
"l_name": "Wonderland"
}
and
{
"f_name": "John",
"id": "1",
"l_name": "McCain"
}
I used your GraphQL schema
Resolvers
Resolver on Query.getChore pointing to the chores table:
{
"version": "2017-02-28",
"operation": "GetItem",
"key": {
"id": $util.dynamodb.toDynamoDBJson($ctx.args.id),
}
}
Resolver on Chore.assignedTo pointing to the users table (note the $ctx.source.assignedTo instead of $ctx.args)
{
"version": "2017-02-28",
"operation": "GetItem",
"key": {
"id": $util.dynamodb.toDynamoDBJson($ctx.source.assignedTo),
}
}
Similarly, resolver on Chore.createdBy pointing to the users table:
{
"version": "2017-02-28",
"operation": "GetItem",
"key": {
"id": $util.dynamodb.toDynamoDBJson($ctx.source.createdBy),
}
}
All resolvers response mapping template use the pass-through.
Running the query
Finally, when running your query:
query getChore {
getChore(id: "36d597c8-2c7e-4f63-93ee-38e5aa8f1d5b") {
id
assignedTo {
id
f_name
l_name
}
createdBy {
id
f_name
l_name
}
}
}
I get the following results:
{
"data": {
"getChore": {
"id": "36d597c8-2c7e-4f63-93ee-38e5aa8f1d5b",
"assignedTo": {
"id": "1",
"f_name": "John",
"l_name": "McCain"
},
"createdBy": {
"id": "2",
"f_name": "Alice",
"l_name": "Wonderland"
}
}
}
}
Hope it helps!
In AWS AppSync, arguments send on the main query don't seem to be forwarded to all children resolvers.
type Query {
article(id: String!, consistentRead: Boolean): Article
book(id: String!, consistentRead: Boolean): Book
}
type Article {
title: String!
id: String!
}
type Book {
articleIds: [String]!
articles: [Article]!
id: String!
}
when I call:
query GetBook {
book(id: 123, consistentRead: true) {
articles {
title
}
}
}
the first query to get the book receives the consistentRead param in $context.arguments, but the subsequent query to retrieve the article does not. ($context.arguments is empty)
I also tried articles(consistentRead: Boolean): [Article]! inside book but no luck.
Does anyone know if it's possible in AppSync to pass arguments to all queries part of the same request?
It is possible to pass arguments from parent to child via the response. Let me explain ...
AppSync has several containers inside $context:
arguments
stash
source
arguments and stash are always cleared before invoking a child resolver as evident from these Cloudwatch logs:
At the very end of the parent execution - arguments and stash data are present.
{
"errors": [],
"mappingTemplateType": "After Mapping",
"path": "[getLatestDeviceState]",
"resolverArn": "arn:aws:appsync:us-east-1:xxx:apis/yyy/types/Query/fields/getLatestDeviceState",
"context": {
"arguments": {
"device": "ddddd"
},
"prev": {
"result": {
"items": [
{
"version": "849",
"device": "ddddd",
"timestamp": "2019-01-29T12:18:34.504+13:00"
}
]
}
},
"stash": {"testKey": "testValue"},
"outErrors": []
},
"fieldInError": false
}
and then at the very beginning of the child resolver - arguments and stash are always blank.
{
"errors": [],
"mappingTemplateType": "Before Mapping",
"path": "[getLatestDeviceState, media]",
"resolverArn": "arn:aws:appsync:us-east-1:yyy:apis/xxx/types/DeviceStatePRODConnection/fields/media",
"context": {
"arguments": {},
"source": {
"items": [
{
"version": "849",
"device": "ddddd",
"timestamp": "2019-01-29T12:18:34.504+13:00"
}
]
},
"stash": {},
"outErrors": []
},
"fieldInError": false
}
Workaround 1 - get the argument from the previous result.
In the example above device is always present in the response of the parent resolver, so I inserted
#set($device = $util.defaultIfNullOrBlank($ctx.args.device, $ctx.source.items[0].device))
into the request mapping template of the child resolver. It will try to get the ID it needs from the arguments and then fall back onto the previous result.
Workaround 2 - add the argument to the parent response
Modify your parent resolver response template to include the arguments:
{
"items": $utils.toJson($context.result.items),
"device": "${ctx.args.device}"
}
and then retrieve it in the request mapping template of the child the same way as in the first workaround.
To achieve availability across all related resolvers (nested or those collection-entity related) for me was fine Workaround 2 (tnx Max for such a good answer) but just for child resolvers.
In another case when I needed to resolve entities from collection query (contains other fields besides entity) property added to response mapping template wasn't available anymore.
So my solution was to set it to request headers:
##Set parent query profile parameter to headers to achieve availability accross related resolvers.
#set( $headers = $context.request.headers )
$util.qr($headers.put("profile", $util.defaultIfNullOrBlank($context.args.profile, "default")))
And read this value from your nested/other request mapping templates:
#set($profile = $ctx.request.headers.profile)
This makes the parent argument available wherever I need it between related resolvers. In your case, it would be 'device' and some default value or without that part if not needed.
Add this to BookQuery Response Mapping Template
#set( $book = $ctx.result )
#set($Articles = []);
#foreach($article in $book.articles)
#set( $newArticle = $article )
$util.qr($newArticle.put("bookID", $book.id))
$util.qr($Articles.add($newArticle))
#end
$util.qr($book.put("articles", $Articles))
$util.toJson($book)
Now, every article will have bookID
You should be able to find consistentRead in $context.info.variables ($context.info.variables.consistentRead):
https://docs.aws.amazon.com/appsync/latest/devguide/resolver-context-reference.html#aws-appsync-resolver-context-reference-info
You don't need to pass arguments to sub-query. Base on your schema and use-case, I think you can adjust your schema like below to have a relationship between Author and Book
type Author {
# parent's id
bookID: ID!
# author id
id: ID!
name: String!
}
type Book {
id: ID!
title: String!
author: [Author]!
}
type Mutation {
insertAuthor(bookID: ID!, id: ID!, name: String!): Author
insertBook(id: ID!, title: String!): Book
}
type Query {
getBook(id: ID!): Book
}
- Create table Author with Author.bookID as a primary key and Author.id as a sort key
- Create table Book with Book.id as a primary key
Then, you have to attach a resolver for Book.author
And here is a resolver for insertAuthor mutation
{
"version" : "2017-02-28",
"operation" : "PutItem",
"key" : {
"bookID" : $util.dynamodb.toDynamoDBJson($ctx.args.bookID),
"id" : $util.dynamodb.toDynamoDBJson($ctx.args.id)
},
"attributeValues" : {
"name" : $util.dynamodb.toDynamoDBJson($ctx.args.name)
}
}
And when you do query getBook you will get a list of author that has the same book id as below
Simply in the child use $ctx.source.id where id is the parameter you need reference from the parent.