Example:
query {
me {
starredPosts {
id
}
}
}
How can the server notice that only the ids are requested, and use the already-fetched user.starredPosts (an array of ids), instead of calling Posts.findOne(id) for each id?
We had the same problem and are in the process of open-sourcing the tools we've built out over the last year and a half internally to address these issues: https://github.com/4Catalyzer/graphql-node-resource/pull/1.
The solution we use is, for object resolvers like that, to resolve them to a "stub" object that contains only the ID, something like:
const childField = {
type: ChildType,
resolve: obj => ({ id: obj.childId }),
};
Then we use DataLoader to fetch the additional fields on the child objects when they're required by using our own default resolver.
We connect to our internal REST API, which supports batching on those requests, so queries that require additional fields get efficiently dispatched and resolved.
However, this does introduce potential for error when writing custom resolvers, as there's no guarantee that obj actually has the relevant fields. We've addressed this by setting up our static types to prevent unchecked access to properties of obj.
You can examine info.fieldNodes[0].selectionSet.selections or use graphql-fields package:
const postsIds = user.starredPosts
const selectionSet = Object.keys(graphqlFields(info))
const onlySelectingId = isEqual(['__typename', 'id'], selectionSet.sort())
if (onlySelectingId) {
return postIds.map(id => ({ id, __typename: 'Post' }))
} else {
return favs.map(id => Post.findOneById(id))
}
Related
What is the best way to collect some specific property from all the leafs of the GraphQL graph, reducing it to some single array? For example, my service functions can "throw" some arbitrary string warnings which I want to collect and supply to the client besides the main data, e.g. expected output:
type EntityOutput {
entity: Entity
warnings: [String!]
}
Resolver:
#Mutation()
async updateEntity(
#Args('id', ParseUUIDPipe) id: string,
#Args('data') input: UpdateDto
): Promise<EntityOutputDto>
{
return {
entity: await this.service.update(id, input),
warnings: [] // ???
};
}
Service method:
async update(id: string, input: UpdateDto): Promise<Entity> {
const entity = await this.repository.findOneOrFail(id, { relations: ['type'] }); // check existence
if (Object.values(input).some(v => v !== undefined)) {
const updateData: Partial<Entity & UpdateDto> = Object.assign({ id }, input);
if (input.isCurrentEntityOfItsType === true) {
await this.typesService.update(entity.type.id, { currentEntityId: id }); // <-- this also can create its own warnings
} else if (input.isCurrentEntityOfItsType === false) {
await this.typesService.update(entity.type.id, { currentEntityId: null as any });
}
await this.repository.save(updateData);
} else {
console.warn(`No properties to change were been provided`); // <-- this is a warning I want to save
}
return this.findOne(id);
}
I think my question can be splitted into 2:
To collect warnings from the service, i.e., in general case, the function calls stack of arbitrary depth. It actually looks more like a general programming problem than a NestJS thing
But even when one implement the feature from the first paragraph the NestJS will walk along the GraphQL graph by itself and there can be additional logs in nested fields.
The solution in its complete general form probably will be over-complicated but at least can anyone suggest the good design for the case represented by the example code?
I have a couple of thoughts:
Should every function in the service return its warnings alongside its main response (for example, in a tuple) so we can incrementally "fold" the array of warnings while "unfolding" the calls stack?
Maybe it would be better to implement using some decorator by which we will mark our service methods?
Maybe RxJS – the NestJS beloved one – can offer us some solution? (I don't know a lot about this library/their philosophy)
Actually the default form of the NestJS output is already looking similar to what I want, it's a JSON with 2 root properties: "errors" and "data". And they can be automatically sent to you simultaneously if the error happened is not so fatal to proceed. Can we somehow overwrite the default response object schema and place warnings there?
The whole question is heavily inspired by this SO discussion but it unfortunately says nothing about the actual possible implementation.
So I've implemented a custom context factory which is executed automatically on every GraphQL request and constructs the object of desired format:
app.module.ts:
export interface AppContext {
warnings: string[];
}
const contextFactory: ContextFunction<any, AppContext> = () => ({
warnings: []
});
Now we can benefit from our newly created interface to add strong typings whenever we reference the context, e.g.:
some.resolver.ts
#Mutation()
async remove(
#Args('id', ParseUUIDPipe) id: string,
#Context() ctx: AppContext
): Promise<FindOneDto>
{
return new FindOneDto(await this.service.remove(id, ctx.warnings));
}
Here the service can add its own warnings to the context.
To collect all of them and return to the API caller I override formatResponse function and append the warnings to the extensions (this is a special GraphQL meta-field serving the developing purposes):
app.module.ts:
const graphqlConfig: GqlModuleOptions = {
context: contextFactory,
formatResponse: (
response: GraphQLResponse | null,
context: GraphQLRequestContext<AppContext>,
): GraphQLResponse =>
{
const warnings = context.context.warnings;
if (warnings.length) {
if (response) {
const extensions = response.extensions || (response.extensions = {});
extensions.warnings = warnings;
} else {
return { extensions: { warnings } };
}
}
return response || {};
},
...
}
Similar approach is used in the official Apollo extension example: https://github.com/apollographql/apollo-server/blob/main/packages/apollo-tracing/src/index.ts.
The only drawback I see now is that injecting the context in resolver's arguments breaks the compliance with auto-generated TypeScript interfaces (I use schema-first approach). In such case, we can switch to per-request-based mode so our resolver/service class instance will be created individually for each and every new request: https://docs.nestjs.com/fundamentals/injection-scopes. Now we can access a context right in the methods without introducing any additional parameters. But this comes with increased latencies and, perhaps, memory-consumption. Another approach will be to create a standalone Nest interceptor.
I've been reading through the graphQL docs and found that they've explained the implementation of the graphql server in 2 ways: one using graphql-yoga which is a fully featured graphql server and another one is using graphql, express-graphql and express. In both cases, we pass the schema and resolver functions while creating the server instance.
But the implementation of resolver function differs. While using graphql-yoga, the resolver function is provided with 4 arguments which contains information about the parent object, arguments received, context, info. whereas in the other case (using graphql), the resolver function only gets the arguments object.
Why is that so ? If I want the info, context objects, how do I get it ?
Using graphql-yoga example: https://graphql.org/learn/execution/
Using graphql example: https://graphql.github.io/graphql-js/mutations-and-input-types/
// Code example using graphql
var express = require('express');
var graphqlHTTP = require('express-graphql');
var { buildSchema } = require('graphql');
var schema = buildSchema(`
type Query {
rollDice(numDice: Int!, numSides: Int): [Int]
}
type Mutation {
addDice(numDice: Int): String
}
`);
var root = {
rollDice({numDice, numSides}) {
return [1, 2];
},
addDice({numDice}) {
console.log("Adding something");
return "Added";
}
};
var app = express();
app.use('/graphql', graphqlHTTP({
schema: schema,
rootValue: root,
graphiql: true,
}));
app.listen(4000);
console.log('Running a GraphQL API server at localhost:4000/graphql');
// Code example using graphql-yoga
let graphqlServer = require("graphql-yoga");
const typeDefs = `
type Query {
rollDice(numDice: Int!, numSides: Int): [Int]
}
type Mutation {
addDice(numDice: Int): String
}
`;
const resolvers = {
Query: {
rollDice(parent, args, context, info) {
console.log(args.numDice);
console.log(args.numSides);
return [1, 2];
}
},
Mutation: {
addDice(parent, args, context, info) {
console.log(args.numDice);
return "Added";
}
}
};
const server = new graphqlServer.GraphQLServer({
typeDefs,
resolvers
});
server.start(() => {
console.log("server started on localhost:4000");
});
Difference between these 2 code snippets:
The resolver functions are present inside appropriate types (i.e. Query, Mutation) in one case. In the other case, they are present inside one root object. This means that I can have methods with same name in Query and Mutation in the first case, whereas in the second case that's not possible since they are keys of a single object and keys should be unique.
Why is this so ? Am I basically missing something ? How can the implementation details differ from one package to another ?
REAL TALK: the GraphQL.js docs are not that great. In my opinion, they never should have used examples with buildSchema in the first place because it understandably leads to this kind of confusion.
GraphQL.js (i.e. the graphql package) is the JavaScript implementation of GraphQL. Building a schema in GraphQL.js is done programmatically, by constructing an instance of the GraphQLSchema class:
const userType = new GraphQLObjectType({
name: 'User',
fields: {
id: {
type: GraphQLID,
},
email: {
type: GraphQLString,
},
},
});
const queryType = new GraphQLObjectType({
name: 'Query',
fields: {
user: {
type: userType,
resolve: () => ({ id: 1, email: 'john.doe#example.com' }),
},
},
});
const schema = new GraphQLSchema({
query: queryType,
})
If we print this schema in Schema Definition Language (SDL), it looks like this:
type Query {
user: User
}
type User {
id: ID
email: String
}
Working with SDL is much easier than having to write out all that code. However, GraphQL.js does not provide a way to build a fully-featured schema from SDL. It does provide a buildSchema function, but this utility constructs a schema without any resolvers (and a number of other features like union/interface type resolution).
The graphql-tools package provides a makeExecutableSchema function that lets you build a schema from SDL and a resolver map object. This is what's used under the hood by apollo-server and graphql-yoga. makeExecutableSchema constructs a schema from SDL using buildSchema and then mutates the resulting object, adding the resolvers in after the fact.
In GraphQL.js, the resolve function (or resolver) for a field takes four parameters -- the parent value, the field's arguments, the context and a GraphQLResolveInfo object. If we're creating a GraphQLObjectType like userType in the above example, this is the optional function we can provide for each of the fields in our object. This is the same function you define when you construct a resolver map to use with graphql-yoga. This is the only implementation of a field resolver.
So what's the deal with buildSchema??
The examples in the docs take advantage of GraphQL's default field resolver:
export const defaultFieldResolver: GraphQLFieldResolver<any, *> = function(
source,
args,
contextValue,
info,
) {
if (typeof source === 'object' || typeof source === 'function') {
const property = source[info.fieldName];
if (typeof property === 'function') {
return source[info.fieldName](args, contextValue, info);
}
return property;
}
};
As you can see, the default resolution logic looks for a property with the same name as the field on the source (parent) value. In our example above, the user resolver returns {id: 1, email: 'john.doe#example.com'} -- this is the value the field resolves to. The field is of the type User. We do not have a resolver defined for our id field, so the default resolver does its thing. The id field resolves to 1 because that's the value of the property named id on the parent object the resolver receives.
However, the parent value can also be a function instead of an object. If it's a function, it gets called first and then the return value is used. What does the function get called with? Well, it can't pass it a parent value (because of infinite recursion), but it can pass it the remaining three parameters (args, context and info). So that's what it does.
Now for the magic trick 🎩🐇
In our example, I can omit the resolver for the user field and pass a function to the root value instead.
const root = {
user: () => ({id: 1, email: 'john.doe#example.com'})
}
The root object is just an optional object that's passed down as the parent value to resolvers at the root level (like your Query or Mutation types). Otherwise, those resolvers would not have a parent value.
Query is an operational root type -- it serves as an "entry point" to the rest of your schema. Any fields on the Query type will be passed the root object as the parent value. If I omit a resolver for the user field, the default resolver will 1) examine the parent object for a property with the same name, 2) find a property and determine that it's a function, 3) call the function, 4) resolve the field to the return value of the function.
TADA!
However, because the function is called by the default resolver, and is not used as a resolver itself, it will only receive the three aforementioned parameters, instead of 4.
This is a neat way to work around not being able to actually provide custom resolvers for a schema, but it's very limited. It only works for root types, so we can't similarly provide fake resolvers for User fields or other types. We can't use interfaces or unions in our schema because we can't provide resolveType functions. And so on...
Hopefully that provides some clarity. And hopefully we can get the docs updated in the near future to avoid all this confusion in the first place.
We are in the situation that the response of our GraphQL Query has to return some dynamic properties of an object. In our case we are not able to predefine all possible properties - so it has to be dynamic.
As we think there are two options to solve it.
const MyType = new GraphQLObjectType({
name: 'SomeType',
fields: {
name: {
type: GraphQLString,
},
elements: {
/*
THIS is our special field which needs to return a dynamic object
*/
},
// ...
},
});
As you can see in the example code is element the property which has to return an object. A response when resolve this could be:
{
name: 'some name',
elements: {
an_unkonwn_key: {
some_nested_field: {
some_other: true,
},
},
another_unknown_prop: 'foo',
},
}
1) Return a "Any-Object"
We could just return any object - so GraphQL do not need to know which fields the Object has. When we tell GraphQL that the field is the type GraphQlObjectType it needs to define fields. Because of this it seems not to be possible to tell GraphQL that someone is just an Object.
Fo this we have changed it like this:
elements: {
type: new GraphQLObjectType({ name: 'elements' });
},
2) We could define dynamic field properties because its in an function
When we define fields as an function we could define our object dynamically. But the field function would need some information (in our case information which would be passed to elements) and we would need to access them to build the field object.
Example:
const MyType = new GraphQLObjectType({
name: 'SomeType',
fields: {
name: {
type: GraphQLString,
},
elements: {
type: new GraphQLObjectType({
name: 'elements',
fields: (argsFromElements) => {
// here we can now access keys from "args"
const fields = {};
argsFromElements.keys.forEach((key) => {
// some logic here ..
fields[someGeneratedProperty] = someGeneratedGraphQLType;
});
return fields;
},
}),
args: {
keys: {
type: new GraphQLList(GraphQLString),
},
},
},
// ...
},
});
This could work but the question would be if there is a way to pass the args and/or resolve object to the fields.
Question
So our question is now: Which way would be recommended in our case in GraphQL and is solution 1 or 2 possible ? Maybe there is another solution ?
Edit
Solution 1 would work when using the ScalarType. Example:
type: new GraphQLScalarType({
name: 'elements',
serialize(value) {
return value;
},
}),
I am not sure if this is a recommended way to solve our situation.
Neither option is really viable:
GraphQL is strongly typed. GraphQL.js doesn't support some kind of any field, and all types defined in your schema must have fields defined. If you look in the docs, fields is a required -- if you try to leave it out, you'll hit an error.
Args are used to resolve queries on a per-request basis. There's no way you can pass them back to your schema. You schema is supposed to be static.
As you suggest, it's possible to accomplish what you're trying to do by rolling your own customer Scalar. I think a simpler solution would be to just use JSON -- you can import a custom scalar for it like this one. Then just have your elements field resolve to a JSON object or array containing the dynamic fields. You could also manipulate the JSON object inside the resolver based on arguments if necessary (if you wanted to limit the fields returned to a subset as defined in the args, for example).
Word of warning: The issue with utilizing JSON, or any custom scalar that includes nested data, is that you're limiting the client's flexibility in requesting what it actually needs. It also results in less helpful errors on the client side -- I'd much rather be told that the field I requested doesn't exist or returned null when I make the request than to find out later down the line the JSON blob I got didn't include a field I expected it to.
One more possible solution could be to declare any such dynamic object as a string. And then pass a stringified version of the object as value to that object from your resolver functions. And then eventually you can parse that string to JSON again to make it again an object on the client side.
I'm not sure if its recommended way or not but I tried to make it work with this approach and it did work smoothly, so I'm sharing it here.
I'm working with GraphQL and having some trouble finding the best way to pipe variables from the query to the result.
I have a schema like so:
type Fragment {
# The id of the fragment
id: String!
# The key of the fragment
key: String!
# The type of component
component_type: String!
# The params used to build the fragment
params: JSON
# Component data
data: JSON
children: [JSON]
items: [JSON]
}
The fragment is meant as a "cms" fragment. I want to pass some query data through to another backend after this resolves.
My query looks like this:
query getFragmentsWithItems($keys: [String!]!
$platform: PlatformType
$version: String
$userInfo: UserInput
$userId: Int
) {
fragmentsWithItems(keys: $keys, platform: $platform, version: $version, userInfo: $userInfo, userId: $userId) {
key
data
children
params
items
}
}
Here's the problem: I have some query data in the data field from the Fragment. That data is not available until that Fragment has resolved. I want to take that data and send it to a different backend. I want to do this with GraphQL, and I was hoping to do something like:
Fragment: () => {
async query(obj, args, context, info, {modles}) => {
const items = await models.getItems(obj.query_string);
}
}
But I need the user_info and user_id that I passed to the original query. Apparently that is only accessible from the info argument which is not meant to be used.
The other path I've taken is to have a manual resolver that does something like so:
const resolveFI = ({ keys, platform, version, userInfo, userId, models }) => {
if (!keys || !keys.length) {
return Promise.resolve(null);
}
return models.release.get({ platform, version }).then(release =>
Promise.all(
keys.map(key =>
models.fragments.get({
key,
platform,
version,
release: release.id
})
)
).then(data => {
const promises = [];
data.rows.forEach(r => {
if (r.data.query_data) {
const d = {
// Can just ignore
filters: r.data.query_data.filters || {},
user_info: userInfo,
user_id: userId
};
promises.push(
new Promise(resolve => {
resolve(
models.itemSearch.get(d).then(i => ({ items: i.items, ...r }))
);
})
);
}
...etc other backends
This works, however a manual promise chain seems to defeat the purpose of using GraphQL.
The last thing I tried was making items a non-scalar type, something like:
type Fragment {
items: ItemSearchResult(user_info: UserInput) etc
But since I can't pipe the actual result from Fragment to the ItemSearchResult that doesn't work.
I realize this is pretty long-winded so I'm open to edits or clarifying.
I'm looking to see if I've missed a better approach or if I should just bag it and have the client apps do the item query after they get the Fragment data back.
It's not that you're not supposed to use info -- it's just a tremendous pain in the butt to use ;) In all seriousness, it's meant to be used for optimization and more advanced use cases, so you shouldn't hesitate to use it if a better solution doesn't present itself. There are libraries out there (like this one) that you can use to parse the object more easily.
That said, there's a couple of ways I imagine you could handle this:
1.) Inside your query resolver(s)
getFragmentsWithItems: async (obj, args, ctx, info) => {
const fragments = await howeverYouDoThat()
const backendCalls = fragments.map(fragment => {
// extract whatever data you need from the fragment
return asyncCallToBackEnd()
})
await backendCalls
return fragments
}
Unfortunately, if you have a lot of different queries returning fragments, you'll end up with redundancy.
2.) Inside the resolver for an existing field (or an additional one) on the Fragment type.
If you go this route, and you need args passed to the query field, you can extract them using the info. Alternatively, you can also mutate the context object inside your query resolver and attach those arguments to it. Then, all resolvers "below" the query resolver (like the resolvers for your Fragment fields) can access those arguments through the context.
3.) Apollo Server lets you define a formatResponse function when configuring its middleware. This essentially provides a hook to do whatever you want with the response before it's returned to the client. You could parse the response inside that function and make the calls to the other backend from there.
I am using graphql-tools. After receiving a GraphQL query, I execute a search using ElasticSearch and return the data.
However, usually the requested query includes only a few of the possible fields, not all. I want to pass only the requested fields to ElasticSearch.
First, I need to get the requested fields.
I can already get the whole query as a string. For example, in the resolver,
const resolvers = {
Query: {
async user(p, args, context) {
//can print query as following
console.log(context.query)
}
.....
}
}
It prints as
query User { user(id:"111") { id name address } }
Is there any way to get the requested fields in a format like
{ id:"", name:"", address:"" }
In graphql-js resolvers expose a fourth argument called resolve info. This field contains more information about the field.
From the GraphQL docs GraphQLObjectType config parameter type definition:
// See below about resolver functions.
type GraphQLFieldResolveFn = (
source?: any,
args?: {[argName: string]: any},
context?: any,
info?: GraphQLResolveInfo
) => any
type GraphQLResolveInfo = {
fieldName: string,
fieldNodes: Array<Field>,
returnType: GraphQLOutputType,
parentType: GraphQLCompositeType,
schema: GraphQLSchema,
fragments: { [fragmentName: string]: FragmentDefinition },
rootValue: any,
operation: OperationDefinition,
variableValues: { [variableName: string]: any },
}
In the fieldNodes field you can search for your field and get the selectionSet for the particular field. From here it gets tricky since the selections can be normal field selections, fragments or inline fragments. You would have to merge all of them to know all fields that are selected on a field.
There is an info object passed as the 4th argument in the resolver. This argument contains the information you're looking for.
It can be helpful to use a library as graphql-fields to help you parse the graphql query data:
const graphqlFields = require('graphql-fields');
const resolvers = {
Query: {
async user(_, args, context, info) {
const topLevelFields = graphqlFields(info);
console.log(Object.keys(topLevelFields)); // ['id', 'name', 'address']
},
};
Similarly for graphql-java you may do the same by extending the field parameters with myGetUsersResolverMethod(... DataFetchingEnvironment env).
This DataFetchingEnvironment would be injected for you and you can traverse through this DataFetchingEnvironment object for any part of the graph/query.
This Object allows you to know more about what is being fetched and what arguments have been provided.
Example:
public List<User> getUsers(final UsersFilter filter, DataFetchingEnvironment env) {
DataFetchingFieldSelectionSet selectionSet = env.getSelectionSet();
selectionSet.getFields(); // <---List of selected fields
selectionSet.getArguments(); // <--- Similarly but MAP
...
}
In fact you may be alluding to look ahead data fetching. The above should give you enough insights into the fields requested and you can take it from there to tailor you downstream calls manually. But also you may look into a more efficient way to do this by using the data fetchers for Building efficient data fetchers by looking ahead