Concurrent asynchronous HTTP requests in Java and/or Spring Boot - java

The issue I need to solve (in Java) is:
Fetch an array of games from an API.
Iterate over games N times:
(async / concurrent)
Make an individual HTTP request for every game in order to get its
details.
Store the details of it in a gamesWithDetails array.
Done. I have my array gamesWithDetails.
I cannot fetch the details of all the games with a single request, I have to hit the API endpoint every time per game. So I want to execute these requests asynchronously from each other.
This is a working example in JavaScript in case it's useful. However I'd like to make it work for Spring Boot.
axios.get(`https://la2.api.riotgames.com/lol/match/v4/matchlists/by-account/${data.accountId}`, {
headers: { "X-Riot-Token": "asdasdasdasdadasdasdasd"}
})
.then(resp => {
const promises = [];
for ( match of resp.data.matches ) {
promises.push(
axios.get(`https://la2.api.riotgames.com/lol/match/v4/matches/${match.gameId}`, {
headers: { "X-Riot-Token": "asdasdasdasdasdasdasdasd"}
})
)
}
Promise.all(promises)
.then(matchesDetails => {
matchesDetails.forEach(({ data }) => console.log(data.gameId));
});
})

Basically you will want to do something like this:
package com.example.demo;
import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.core.type.TypeReference;
import com.fasterxml.jackson.databind.ObjectMapper;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.web.client.RestTemplate;
import java.util.List;
import java.util.Map;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.ExecutionException;
import java.util.concurrent.Executor;
import java.util.concurrent.Executors;
import java.util.stream.Collectors;
public class GamesProcessor {
private static final String GAME_URI_BASE = "https://la2.api.riotgames.com/lol/match/v4/matches/";
private static final String ACCOUNT_URI_BASE = "https://la2.api.riotgames.com/lol/match/v4/matchlists/by-account/";
private Executor executor = Executors.newFixedThreadPool(Runtime.getRuntime().availableProcessors() - 1);
#Autowired
private RestTemplate restTemplate;
public void processGames(String accountId) throws JsonProcessingException, ExecutionException, InterruptedException {
String responseAsString = restTemplate.getForObject(ACCOUNT_URI_BASE + accountId, String.class);
ObjectMapper objectMapper = new ObjectMapper();
if (responseAsString != null) {
Map<String, Object> response = objectMapper.readValue(responseAsString, new TypeReference<Map<String, Object>>() {
});
List<Map<String, Object>> matches = (List<Map<String, Object>>) ((Map<String, Object>) response.get("data")).get("matches");
List<CompletableFuture<Void>> futures = matches.stream()
.map(m -> (String) m.get("gameId"))
.map(gameId -> CompletableFuture.supplyAsync(() -> restTemplate.getForObject(GAME_URI_BASE + gameId, String.class), executor)
.thenAccept(r -> {
System.out.println(r); //do whatever you wish with the response here
}))
.collect(Collectors.toList());
// now we execute all requests asynchronously
CompletableFuture.allOf(futures.toArray(new CompletableFuture[0])).get();
}
}
}
Please note that it is not a refined code, but just a quick example of how to achieve this. Ideally you would replace that JSON processing that I've done "by hand" using Map by a response bean that matches the structure of the response you get from the service you are calling.
A quick walk through:
String responseAsString = restTemplate.getForObject(ACCOUNT_URI_BASE + accountId, String.class);
This executes the first REST request and gets it as a String (the JSON response). You will want to properly map this using a Bean object instead. Then this is processed using the ObjectMapper provided by Jackson and transformed into a map so you can navigate the JSON and get the matches.
List<CompletableFuture<Void>> futures = matches.stream()
.map(m -> (String) m.get("gameId"))
.map(gameId -> CompletableFuture.supplyAsync(() -> restTemplate.getForObject(GAME_URI_BASE + gameId, String.class), executor)
.thenAccept(r -> {
System.out.println(r); //do whatever you wish with the response here
}))
.collect(Collectors.toList());
Once we have all the matches we will use the Stream API to transform them into CompletableFutures that will be executed asynchronously. Each thread will make another request in order to get the response for each individual matchId.
System.out.println(r);
This will be executed for each response that you get for each matchId, just like in your example. This should also be replaced by a proper bean matching the output for clearer processing.
Note that List<CompletableFuture<Void>> futures only "holds the code" but will not get executed until we combine everything in the end using CompletableFuture.allOf(futures.toArray(new CompletableFuture[0])).get(); and execute the blocking get() method.

Quite interesting question as JavaScript implements the famous event loop which means its functions are asynchronous and non-blocking. Spring Boot restTemplate class will block the thread of execution until the response is back, therefore wasting a lot of resources (the one thread per request model).
#Slacky's answer is technically right as you asked about asynchronous HTTP requests but I'd like to share a better option which is both asynchronous and non-blocking, meaning a single thread is able to handle 100s or even 1000s of requests and their responses (reactive programming).
The way to implement in Spring Boot the equivalent to your JavaScript example is to use the Project Reactor WebClient class which is a non-blocking, reactive client to perform HTTP requests.
It is also worth mentioning that Java being statically typed requires you to declare classes to represent your data, in this case something like (using Lombok for brevity):
#Data
class Match {
private String gameId;
// ...
}
#Data
class MatchDetails {
// ...
}
Here is the code following #Slacky's answer naming convention to make the comparison easier.
public class GamesProcessor {
private static final String BASE_URL = "https://la2.api.riotgames.com";
private static final String GAME_URI = "/lol/match/v4/matches/%s";
private static final String ACCOUNT_URI = "/lol/match/v4/matchlists/by-account/%s";
public static List<MatchDetails> processGames(String accountId) {
final WebClient webClient = WebClient
.builder()
.baseUrl(BASE_URL)
.defaultHeader("X-Riot-Token", "asdasdasdasdadasdasdasd")
.build();
// Issues the first request to get list of matches
List<Match> matches = webClient
.get()
.uri(String.format(ACCOUNT_URI, accountId))
.accept(MediaType.APPLICATION_JSON)
.retrieve()
.bodyToMono(new ParameterizedTypeReference<List<Match>>() {})
.block(); // blocks to wait for response
// Processes the list of matches asynchronously and collect all responses in a list of matches details
return Flux.fromIterable(matches)
.flatMap(match -> webClient
.get()
.uri(String.format(GAME_URI, match.getGameId()))
.accept(MediaType.APPLICATION_JSON)
.retrieve()
.bodyToMono(MatchDetails.class))
.collectList()
.block(); // Blocks to wait for all responses
}
}

Related

Webflux returning Mono and using info in Object in Mono to set header

Let's say my Webflux handler returns a Mono on a product creation
That's easy to do.
But now, I want to complete the response with a location in the header.
To do so, I need to get the created product ID.
In my example, I used a block() which fails the reactive idea of the handler.
public Mono<ServerResponse> handleRequest(ServerRequest serverRequest) {
...
Mono<Product> monoProduct = // Service call to get the Mono<Product>
return ServerResponse.ok().contentType(MediaType.APPLICATION_JSON)
.location(URI.create(String.format("/api/products/%s",
monoProduct.block().getId()))))
.body(monoProduct), ProductResponse.class);
}
How can I perform such a task without breaking the reactive principles?
You don't really need to block. You need to build reactive flow combining different operators.
In your case it could look like
public Mono<ServerResponse> handleRequest(ServerRequest serverRequest) {
return getProduct() // Service call to get the Mono<Product>
.map(product -> mapToResponse(product)) // Product -> ProductResponse
.flatMap(response ->
ServerResponse.ok()
.contentType(MediaType.APPLICATION_JSON)
.location(URI.create(String.format("/api/products/%s", response.getId())))
.body(BodyInserters.fromValue(response))
);
}

Filtering response using webclient

I'm new using webclient to cosume a Rest API and I want to know how can I filter a response to match only what I want.
So I have this endpoint which brings me a customerById but I need to show only the the systemsId = 100
Let me show:
#GetMapping("/getCucoId/{cucoId}")
public Mono<CuCoPerson> getCucoRelationById(#PathVariable Integer cucoId) {
WebClient webClient = WebClient.create();
return webClient.get()
.uri(GET_RELATION_BY_ID + cucoId)
.header("Accept", "application/json")
.header("Authorization", "Bearer eyJraWQiOi.....")
.retrieve()
.bodyToMono(CuCoPerson.class);
}
And the POJO:
#Data
#NoArgsConstructor
#AllArgsConstructor
public class CuCoPerson {
private Integer cucoID;
private List<CustomerRelation> relatedCustomers;
}
And this is the response in Postman:
{
"cucoID": 100288298,
"relatedCustomers": [
{
"customerId": "F6305957",
"systemId": 100
},
{
"customerId": "F8364917",
"systemId": 400
},
{
"customerId": "F4194868",
"systemId": 101
}
]
}
So I need only to show the relatedCustomers who only have a systemID = 100
Thanks in advance!
I think you are looking for either Mono.map(), or Mono.flatMap().
Remove the customers, who don't match in the function, and return the changed object.
#GetMapping("/getCucoId/{cucoId}")
public Mono<CuCoPerson> getCucoRelationById(#PathVariable Integer cucoId) {
WebClient webClient = WebClient.create();
return webClient.get()
.uri(GET_RELATION_BY_ID + cucoId)
.header("Accept", "application/json")
.header("Authorization", "Bearer eyJraWQiOi.....")
.retrieve()
.bodyToMono(CuCoPerson.class)
.map(cuCoPerson -> {
List<CustomerRelation> matches = cuCoPerson.getRelatedCustomers()
.stream()
.filter(relation -> relation.getSystemId().equals(100))
.collect(Collectors.toList());
cuCoPerson.setRelatedCustomers(matches);
return cuCoPerson;
});
}
As mentioned in other answers, doing this kind of filtering on client side is bad practice. If possible the API should expose parameter for systemId and return only the data you need.
First of all this is bad Backend design. Since you (client side) have the need for such info you should have an endpoint available to get your info by customer ID and System ID. Data filtering should be done on the backend and the client should remain as thin as possible. Otherwise you perform logic both on server-side and client-side plus you send some completely unneeded info over the network and then filter it out on the client side. But if that's your available resources and you can't change the server-side then you you take your JSON response, and parse it into your class that maps to that JSON or just into Map<String,Object> if you don't have the class for that JSON and filter it out yourself. To parse a json string to map or a specific POJO you can use Jackson library - method readValue() of ObjectMapper Or Gson library. Also, I wrote my own simple wrapper over Jackson library that simplifies the use. Class JsonUtils is available as part of MgntUtils library (written and maintained by me). This class just has the methods that allow you to parse and de-serialize from/to Json String from/to a class instance. Here is a simple example on how to parse your JSON string into a Map<String,Object> with JSONUtils class
Map<String,Object> myMap
try {
myMap = JsonUtils.readObjectFromJsonString(myJsonStr, Map.class);
} catch(IOException ioe) {
....
}
MgntUtils library available as Maven artifact and on Github (including source code and Javadoc)
it could be something like this:
return webClient.get()
.uri("/someUrl")
.header("Accept", "application/json")
.header("Authorization", "Bearer eyJraWQiOi.....")
.retrieve()
.bodyToFlux(CuCoPerson.class)
.filter(cuCoPerson -> cuCoPerson.getRelatedCustomers().stream().anyMatch(cr -> cr.getSystemId() == 100))
.take(1)
.next();
But this has disadvantage that you are filtering the results on client side (your application). I would try to ask the api provider if s/he is able to provide filter parameter (something like systemId) and in that case you would call the endpoint with that query param.

How to wait for subscribe to finish?

I want to make an asynchronous rest call for which I'm using spring webclient and getting back a Mono. I'm also doing some database calls in parallel but it can't be done reactively due to some reason.
Map<String, Object> models = new HashMap<>();
Mono<User> users = this.webClient...;
users.map(resp -> new UserState(userRequest, resp))
.subscribe(response -> {
models.put("userState", response);
});
Iterable<Product> messages = this.productRepository.findAll();
models.put("products", messages);
//Wait for users.subscribe to finish <<<<<<<<<<<<<HERE
return new ModelAndView("messages/list", models);
How do I wait for subscribe to finish before returning ModelAndView. This would have been easy if I was using a Future where I can do get() whenever I want.
You can wrap the blocking call in a Mono executed on a separate scheduler, zip it with the Mono containing UserState data and transform their combination into a Mono<ModelAndView> (which can be returned from Spring controller methods). The calls will be executed in parallel, results will be combined when both calls are completed.
You can define a single bounded scheduler per application specifically for blocking calls and provide it as a constructor argument to any class that makes blocking calls.
The code will look as follows:
#Configuration
class SchedulersConfig {
#Bean
Scheduler parallelScheduler(#Value("${blocking-thread-pool-size}") int threadsCount) {
return Schedulers.parallel(threadsCount);
}
}
#RestController
class Controller {
final Scheduler parallelScheduler;
...
Mono<User> userResponse = // webClient...
Mono<Iterable<Product>> productsResponse = Mono.fromSupplier(productRepository::findAll)
.subscribeOn(parallelScheduler);
return Mono.zip(userResponse, productsResponse, (user, products) ->
new ModelAndView("messages/list",
ImmutableMap.of(
"userState", new UserState(userRequest, user),
"products", products
))
);
}
Update based on the comment:
If you just need to execute HTTP call asynchronously and then join it with the database results you can do the following
Map<String, Object> models = new HashMap<>();
Mono<User> userMono = webClient...;
CompletableFuture<User> userFuture = userMono.toFuture();
Iterable<Product> messages = productRepository.findAll();
User user = userFuture.join();
models.put("products", messages);
models.put("userState", new UserState(userRequest, user));
return new ModelAndView("messages/list", models);

Java Parallel HTTP requests with CompleteableFuture not very performant

I have a web service that makes http calls to another service. The web service breaks down one-to-many requests and attempts to make parallel one-to-one requests. For testing performance, I have kept the throughput to the backend constant. For example, I was able to achieve a throughput of 1000 req/sec with a 99th percentile latency of 100ms. So to test parallel requests that get broken down to 2 requests to the backend per each request to the web service, I sent 500 req/sec but achieved only a 150ms 99th percentile latency. Am I creating thread contention and/or making blocking http calls with the following code?
import java.util.HashMap;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.ConcurrentMap;
import java.util.stream.Collectors;
public class Foo {
private HTTPClient myHTTPClient = new HTTPClient("http://my_host.com"); //java ws rs http client
private interface Handler<REQ, RES> {
RES work(REQ req);
}
private <REQ, RES> CompletableFuture<RES> getAsync(REQ req, Handler<REQ, RES> handler) {
CompletableFuture<RES> future = CompletableFuture.supplyAsync(() -> {
return handler.work(req);
});
return future;
}
public RouteCostResponse getRouteCost(Point sources, List<Point> destinations) {
Map<String, Request> requests = new HashMap<>();
// create request bodies and keep track of request id's
for (Point destination : destinations) {
requests.put(destination.getId(), new RouteCostRequest(source, destination))
}
//create futures
ConcurrentMap<String, CompletableFuture<RouteCost>> futures = requests.entrySet().parallelStream()
.collect(Collectors.toConcurrentMap(
entry -> entry.getKey(),
entry -> getAsync(entry.getValue(), route -> myHTTPClient.getRoute(route)))
));
//retrieve results
ConcurrentMap<String, RouteCost> result = futures.entrySet().parallelStream()
.collect(Collectors.toConcurrentMap(
entry -> entry.getKey(),
entry -> entry.getValue().join()
));
RouteCostResponse response = new RouteCostResponse(result);
return response;
}
}
There is no thread contention with the following code, though it seems i have run into I/O issues. The key is to use an explicit thread pool. ForkJoinPool or Executors.fixedThreadPool
import java.util.HashMap;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.ConcurrentMap;
import java.util.concurrent.ForkJoinPool;
import java.util.stream.Collectors;
public class Foo {
private HTTPClient myHTTPClient = new HTTPClient("http://my_host.com"); //java ws rs http client
private static final ForkJoinPool pool = new ForkJoinPool(1000);
private interface Handler<REQ, RES> {
RES work(REQ req);
}
private <REQ, RES> CompletableFuture<RES> getAsync(REQ req, Handler<REQ, RES> handler) {
CompletableFuture<RES> future = CompletableFuture.supplyAsync(() -> {
return handler.work(req);
});
return future;
}
public RouteCostResponse getRouteCost(Point sources, List<Point> destinations) {
Map<String, Request> requests = new HashMap<>();
// create request bodies and keep track of request id's
for (Point destination : destinations) {
requests.put(destination.getId(), new RouteCostRequest(source, destination))
}
//create futures
ConcurrentMap<String, CompletableFuture<RouteCost>> futures = requests.entrySet().stream()
.collect(Collectors.toConcurrentMap(
entry -> entry.getKey(),
entry -> getAsync(entry.getValue(), route -> myHTTPClient.getRoute(route)))
));
//retrieve results
ConcurrentMap<String, RouteCost> result = futures.entrySet().stream()
.collect(Collectors.toConcurrentMap(
entry -> entry.getKey(),
entry -> entry.getValue().join()
));
RouteCostResponse response = new RouteCostResponse(result);
return response;
}
}

How to end chained http requests in RxJava Vert.x?

How to end chained requests in Rx Vert.X ?
HttpClient client = Vertx.vertx().createHttpClient();
HttpClientRequest request = client.request(HttpMethod.POST,
"someURL")
.putHeader("content-type", "application/x-www-form-urlencoded")
.putHeader("content-length", Integer.toString(jsonData.length())).write(jsonData);
request.toObservable().
//flatmap HttpClientResponse -> Observable<Buffer>
flatMap(httpClientResponse -> { //something
return httpClientResponse.toObservable();
}).
map(buffer -> {return buffer.toString()}).
//flatmap data -> Observable<HttpClientResponse>
flatMap(postData -> client.request(HttpMethod.POST,
someURL")
.putHeader("content-type", "application/x-www-form-urlencoded")
.putHeader("content-length", Integer.toString(postData.length())).write(postData).toObservable()).
//flatmap HttpClientResponse -> Observable<Buffer>
flatMap(httpClientResponse -> {
return httpClientResponse.toObservable();
})......//other operators
request.end();
Notice that I have .end() for the top request. How do I end request that is inside of the .flatmap ? Do I even need to end it ?
There are multiple ways to ensure to call request.end(). But I would dig into documentation of Vert.x or just open source code if there is one, to see if it does call end() for you. Otherwise one could be
final HttpClientRequest request = ...
request.toObservable()
.doOnUnsubscribe(new Action0() {
#Override
public void call() {
request.end();
}
});
I think you can do something like the following code.
The main idea is that you don't directly use the HttpClientRequest as obtained by the Vertx client. Instead you create another flowable that will invoke end() as soon as the first subscription is received.
Here, for instance, you can obtain the request through a pair custom methods: in this case request1() and request2(). They both use doOnSubscribe() to trigger the end() you need. Read its description on the ReactiveX page.
This examle uses vertx and reactivex, I hope you could use this set up.
import io.reactivex.Flowable;
import io.vertx.core.http.HttpMethod;
import io.vertx.reactivex.core.Vertx;
import io.vertx.reactivex.core.buffer.Buffer;
import io.vertx.reactivex.core.http.HttpClient;
import io.vertx.reactivex.core.http.HttpClientRequest;
import io.vertx.reactivex.core.http.HttpClientResponse;
import org.junit.Test;
public class StackOverflow {
#Test public void test(){
Buffer jsonData = Buffer.buffer("..."); // the json data.
HttpClient client = Vertx.vertx().createHttpClient(); // the vertx client.
request1(client)
.flatMap(httpClientResponse -> httpClientResponse.toFlowable())
.map(buffer -> buffer.toString())
.flatMap(postData -> request2(client, postData) )
.forEach( httpResponse -> {
// do something with returned data);
});
}
private Flowable<HttpClientResponse> request1(HttpClient client) {
HttpClientRequest request = client.request(HttpMethod.POST,"someURL");
return request
.toFlowable()
.doOnSubscribe( subscription -> request.end() );
}
private Flowable<HttpClientResponse> request2(HttpClient client, String postData) {
HttpClientRequest request = client.request(HttpMethod.POST,"someURL");
// do something with postData
return request
.toFlowable()
.doOnSubscribe( subscription -> request.end() );
}
}

Categories

Resources