Get access to DataFetchingEnvironment from KotlinDataLoader

[JajaComp]

Hi! Can i get access to DataFetchingEnvironment from KotlinDataLoader? Or how i can add arguments to subRequests?

[smyrick]

KotlinDataLoader is an abstraction on graphql-java DataLoader that the spring server library will autowire for you.

You can always manually create and load data loaders where you can give it info in the DataFetchingEnvironment. Infact during execution, you have to use the DataFetchingEnvironment to get data loaders in the first place.

See the example app:

graphql-kotlin/examples/server/spring-server/src/main/kotlin/com/expediagroup/graphql/examples/server/spring/query/DataLoaderQuery.kt

Lines 47 to 57 in 2a498e8

	/**
	* Register a custom [DataFetcher] to get a [Company] any time it appears in the schema
	*/
	@Component("CompanyDataFetcher")
	class CompanyDataFetcher : DataFetcher<CompletableFuture<Company>> {
	override fun get(environment: DataFetchingEnvironment): CompletableFuture<Company> {
	val companyId = environment.getSource<Employee>().companyId
	return environment.getValueFromDataLoader(CompanyDataLoader.name, companyId)
	}
	}

You can get more info from the data loader docs directly: https://github.com/graphql-java/java-dataloader

[JajaComp]

I am having a problem understanding architecture. With parameters, everything is simple, but how to filter out unnecessary data? At first I thought that it was possible to do this using DataLoader, but as I understand it, in this case it is necessary to refuse and use a single query and get all the people from the database myself, then manually get the cities of these users (Or use join), manually fill in the structure the returned data, through the Context, prevent the use of DataLoader Or is there a better way?

[smyrick]

You could fetch all people from your DB if you wanted to do extra logic that a DB search can't handle and then do the filter yourself, that would be an option but maybe not the best

fun person(city: String?): List<Person> {
  val allPeople: List<People> = getAllPeopleFromDB()
  
  return if (city == null) {
    allPeople
  } else {
    allPeople.filter { person -> person.city == city }
  }
}

But it would probably be better to form a DB query based off the API input then run that. No need for data loaders or context

const val query = "SELECT * FROM people WHERE city = ?"

fun person(city: String?): List<Person> {
  return if (city == null) {
    getAllPeopleFromDB()
  } else {    
    val dbQuery = buildQuery(query, city)
    return runQuery(dbQuery)
  }
}

The point is that all this is hidden from API users and all they do is this

query {
  person(city = "New York") {
   name
  }
}

[smyrick]

All of this is separate from GraphQL. This is part of schema design and the whole point of having an API on top of your DB is to hide implementation details.

Try to think of what use cases your API clients are doing and design the schema to serve them best. Then it is the server side job to handle all the technical details. You probably don't want to just expose everything your DB can do directly

[JajaComp]

And i can't use custom logic in DataLoader for exclude useless data like in my sample? It is important for me to understand exactly the possible logic in subqueries.

[smyrick]

To implement the schema you shared above you should have a top level kotlin function called person

fun person(input: WhatEverYouWant): List<Person> { 
  TODO("implement me")
}

In this function you need to some how get the list of people. You can do that by calling your database, calling another API, generate it randomly or any other option. If you want to use DataLoader, it is just a cache on top of your call and it returns the objects.

Once you have your list of people you can write addition server-side code to modify results anyway you want based on headers, query input, or the response from your calls.

It sounds like you may be misinterpreting how GraphQL in general works and where this library and the schema generator fits in to your tech stack. I recommend you first check out some of our example videos: https://expediagroup.github.io/graphql-kotlin/docs/blogs-and-videos