getData() async {
http.Response response = await http.get('https://www.example.com/);
print(response.body);
}
The above function works to get the HTML code of a page but it fails in some cases. The function is sometimes never completed and it waits forever to get response( For example, if the app is opened while internet is off and even when its turned on, it never connects). In such situations is there any way to retry ?
I tried the http retry package but it gives me 15+ errors.
Example code for how this could be done:
import 'package:http/http.dart' as http;
import 'dart:convert';
Future<List> loadData() async {
bool loadRemoteDatatSucceed = false;
var data;
try {
http.Response response = await http.post("https://www.example.com",
body: <String, String>{"username": "test"});
data = json.decode(response.body);
if (data.containsKey("success")) {
loadRemoteDatatSucceed = true;
}
} catch (e) {
if (loadRemoteDatatSucceed == false) retryFuture(loadData, 2000);
}
return data;
}
retryFuture(future, delay) {
Future.delayed(Duration(milliseconds: delay), () {
future();
});
}
You can use RetryPolicy from http package to retry your connection, just create your own class and inherit form RetryPolicy and override these function like the following example, then create a Client using HttpClientWithInterceptor.build and add your custom retryPolicy as a parameter, this will retry your request for a number of times until a condition is met, if not, it'll just stop retrying.
import 'package:http/http.dart';
class MyRetryPolicy extends RetryPolicy {
final url = 'https://www.example.com/';
#override
// how many times you want to retry your request.
int maxRetryAttempts = 5;
#override
Future<bool> shouldAttemptRetryOnResponse(ResponseData response) async {
//You can check if you got your response after certain timeout,
//or if you want to retry your request based on the status code,
//usually this is used for refreshing your expired token but you can check for what ever you want
//your should write a condition here so it won't execute this code on every request
//for example if(response == null)
// a very basic solution is that you can check
// for internet connection, for example
try {
final result = await InternetAddress.lookup('google.com');
if (result.isNotEmpty && result[0].rawAddress.isNotEmpty) {
return true;
}
return false;
} on SocketException catch (_) {
return false;
}
}
}
then create and use a client to make your requests.
it will automatically retry the request if the condition you wrote is met.
Client client = HttpClientWithInterceptor.build(
retryPolicy: ExpiredTokenRetryPolicy(),
);
final response = await client.get('https://www.example.com/);
there is also a package to check for internet connection if that your problem, see connectivity
You can use try-catch blocks inside async functions like you would in synchronous code. Perhaps you'd be able to add some sort of error handling mechanism in the function, and retry the function on error? Here's some documentation on that one.
Example from the docs:
try {
var order = await getUserOrder();
print('Awaiting user order...');
} catch (err) {
print('Caught error: $err');
}
You can also catch specific Exceptions, per this github issue.
doLogin(String username, String password) async {
try {
var user = await api.login(username, password);
_view.onLoginSuccess(user);
} on Exception catch(error) {
_view.onLoginError(error.toString());
}
}
EDIT: This may also help.
While we're at it, look here for a function that reattempts an async operation however many times you need.
Related
I'm a beginner developer and I have problem with implementation of BloC framework. Let's assume that I have this code (Model, NetworkService, Repository, Cubit, State, Widget):
class NetworkService {
Future getData(Uri uri) async {
try {
http.Response httpsResponse = await http.get(
uri,
headers: {
// some headers //
},
);
if (httpsResponse.statusCode == 200) {
return httpsResponse.body;
} else {
throw 'Request failed with status: ${httpsResponse.statusCode}';
}
} catch (e) {
// What I shloud return here?
return e.toString();
}
}
Future<List<dynamic>> fetchData() async {
final uri = Uri.parse('some url');
var data = await getData(uri);
return = jsonDecode(data) as List;
}
}
class Repository {
final NetworkService networkService = NetworkService();
Future<List<SomeObject>> fetchDataList() async {
final dataRaw =
await networkService.fetchDataList();
return dataRaw.map((e) => SomeObject.fromJson(e)).toList();
}
}
class SomeCubit extends Cubit<CubitState> {
final Repository repository;
SomeCubit(this.repository) : super(LoadingState()) {
fetchDataList();
}
void fetchDataList() {
try {
repository
.fetchDataList()
.then((dataList) => emit(LoadedState(dataList)));
} catch (e) {
// What I shloud return here?
emit(ErrorState(e.toString()));
}
}
}
How to make this code "bullet proof" because I don't know how to "pass" error from NetworkService to Cubit? It works fine till I have dynamic responses in functions but in Repository class I want to return List of specific objects and when function fail I will return null. If I write try/catch I have to provide return statement in catch block - and I can't return List. I want to return some kind of Error...
I suggest that you use the excellent class named Either from the dartz package. It will allow you to return X if things went bad, and return Y if all is well, as such: Future<Either<X, Y>>
Then you can check on your variable (e.g. result) as follows: result.isLeft() for error, or do result.fold( ... ) to easily handle the return type (error or success).
In your particular case you could do as follows when returning from the repository to the cubit:
Future<Either<RepositoryError, List<SomeObject>>> fetchDataList() async { ... }
Where RepositoryError could be a class containing information about the type of error.
So in the cubit you do:
final result = await repository.fetchDataList();
emit(
result.fold(
(error) => ErrorState(error),
(dataList) => LoadedState(dataList)
)
);
Then you continue with this pattern all the way to NetworkService getData(). Either with the same common "error class" in the Repository and the NetworkService, or separate ones in the different layers and you "translate" between different "error classes". Perhaps it makes sense to have a NetworkServiceError that is returned there..
In your NetworkService you could do as follows:
Future<Either<NetworkServiceError, String>> getData(Uri uri) async { ... }
Future<Either<NetworkServiceError, List<dynamic>>> fetchData() async { ... }
This will give you great flexibility and passing of information from the service, to the repository and to the cubit.
You can let exceptions propagate through Futures from NetworkService up to the cubit, by removing the try/catch from getData.
I have a custom http client that intercepts responses to provide service unavailable events via a stream:
class MyHttpClient extends http.BaseClient {
final String apiBase;
MyHttpClient({required this.apiBase});
final _controller = StreamController<ServiceUnavailableEvent>.broadcast();
Stream<ServiceUnavailableEvent> get stream => _controller.stream;
#override
Future<http.StreamedResponse> send(http.BaseRequest request) async {
final response = await request.send();
if (request.url.toString().startsWith(apiBase) &&
response.statusCode == HttpStatus.serviceUnavailable) {
_controller.sink.add(ServiceUnavailableEvent(isAvailable: false));
} else {
_controller.sink.add(ServiceUnavailableEvent(isAvailable: true));
}
return response;
}
Future<void> close() async {
await _controller.close();
}
}
The client is instanciated once in the main flutter function:
void main() {
final client = MyHttpClient(apiBase: apiBase)
runApp(MyApp(client: client));
}
Everything works fine and I added the close method on the custom http client that closes the stream. But when/how can I call this close method ?
I thought about AppLifeCycleEvent but all state seem to be the wrong place, because I only want to close the client when the app really shuts down (i.e. if the user re-opens/resumes the app a new client must be created in order to recreate the subscription).
how do you know it is still open when app shutdown. i think platform(ios/android) close it.
I'm getting this error in my StateNotifiers when one hits the back button on their webpage. I've isolated it to happening where the longRunningAPI request is below.
Exception has occurred.
"Error: Bad state: Tried to use RunListNotifier after `dispose` was called.
and I have code like this.
final runListController = StateNotifierProvider.autoDispose
.family<RunListNotifier, AsyncValue<List<Run>>, RunListParameter>(
(ref, param) {
return RunListNotifier(read: ref.read, param: param);
});
class RunListNotifier extends StateNotifier<AsyncValue<List<Run>>> {
RunListNotifier({required this.read, required this.param})
: super(AsyncLoading()) {
fetchViaAPI(param);
}
final Reader read;
final RunListParameter param;
void fetchViaAPI(RunListParameter param) async {
state = AsyncLoading();
try {
List<Run> stuff = await read(apiProvider).longRunningAPI(param: param);
state = AsyncData(stuff);
} catch (e) {
state = AsyncError(e);
}
}
}
is it safe to simply do something like this in the catch?
} catch (e) {
if (e.runtimeType.toString() == 'StateError') {
// ignore the error
} else {
state = AsyncError(e);
}
}
I believe you could solve this problem by checking mounted before setting the state after your API call like so:
List<Run> stuff = await read(apiProvider).longRunningAPI(param: param);
if (!mounted) return;
state = AsyncData(stuff);
This simply checks if dispose was called and if so, don't attempt to modify the state.
Another resource that could be useful is adding a cancelToken to your API call and canceling if the provider is disposed.
final longRunningApi = FutureProvider.autoDispose.family<List<Run>, RunListParameter>((ref, param) async {
final cancelToken = CancelToken();
ref.onDispose(cancelToken.cancel);
final api = await ref.watch(apiProvider);
final res = await api.longRunningApi(param, cancelToken);
ref.maintainState = true;
return res;
});
Then you'd have to add the cancelToken to your actual request. A great example of this in the marvel example project by the author of Riverpod can be found here.
This method submits a simple HTTP request and calls a success or error callback just fine:
void _getSimpleReply( String command, callback, errorCallback ) async {
try {
HttpClientRequest request = await _myClient.get( _serverIPAddress, _serverPort, '/' );
HttpClientResponse response = await request.close();
response.transform( utf8.decoder ).listen( (onData) { callback( onData ); } );
} on SocketException catch( e ) {
errorCallback( e.toString() );
}
}
If the server isn't running, the Android-app more or less instantly calls the errorCallback.
On iOS, the errorCallback takes a very long period of time - more than 20 seconds - until any callback gets called.
May I set for HttpClient() a maximum number of seconds to wait for the server side to return a reply - if any?
There are two different ways to configure this behavior in Dart
Set a per request timeout
You can set a timeout on any Future using the Future.timeout method. This will short-circuit after the given duration has elapsed by throwing a TimeoutException.
try {
final request = await client.get(...);
final response = await request.close()
.timeout(const Duration(seconds: 2));
// rest of the code
...
} on TimeoutException catch (_) {
// A timeout occurred.
} on SocketException catch (_) {
// Other exception
}
Set a timeout on HttpClient
You can also set a timeout on the HttpClient itself using HttpClient.connectionTimeout. This will apply to all requests made by the same client, after the timeout was set. When a request exceeds this timeout, a SocketException is thrown.
final client = new HttpClient();
client.connectionTimeout = const Duration(seconds: 5);
You can use timeout
http.get(Uri.parse('url')).timeout(
const Duration(seconds: 1),
onTimeout: () {
// Time has run out, do what you wanted to do.
return http.Response('Error', 408); // Request Timeout response status code
},
);
The HttpClient.connectionTimeout didn't work for me. However, I knew that the Dio packet allows request cancellation. Then, I dig into the packet to find out how they achieve it and I adapted it to me. What I did was to create two futures:
A Future.delayed where I set the duration of the timeout.
The HTTP request.
Then, I passed the two futures to a Future.any which returns the result of the first future to complete and the results of all the other futures are discarded. Therefore, if the timeout future completes first, your connection times out and no response will arrive. You can check it out in the following code:
Future<Response> get(
String url, {
Duration timeout = Duration(seconds: 30),
}) async {
final request = Request('GET', Uri.parse(url))..followRedirects = false;
headers.forEach((key, value) {
request.headers[key] = value;
});
final Completer _completer = Completer();
/// Fake timeout by forcing the request future to complete if the duration
/// ends before the response arrives.
Future.delayed(timeout, () => _completer.complete());
final response = await Response.fromStream(await listenCancelForAsyncTask(
_completer,
Future(() {
return _getClient().send(request);
}),
));
}
Future<T> listenCancelForAsyncTask<T>(
Completer completer,
Future<T> future,
) {
/// Returns the first future of the futures list to complete. Therefore,
/// if the first future is the timeout, the http response will not arrive
/// and it is possible to handle the timeout.
return Future.any([
if (completer != null) completeFuture(completer),
future,
]);
}
Future<T> completeFuture<T>(Completer completer) async {
await completer.future;
throw TimeoutException('TimeoutError');
}
This is an example of how to extend the http.BaseClient class to support timeout and ignore the exception of the S.O. if the client's timeout is reached first.
you just need to override the "send" method...
the timeout should be passed as a parameter to the class constructor.
import 'dart:async';
import 'package:http/http.dart' as http;
// as dart does not support tuples i create an Either class
class _Either<L, R> {
final L? left;
final R? right;
_Either(this.left, this.right);
_Either.Left(L this.left) : right = null;
_Either.Right(R this.right) : left = null;
}
class TimeoutClient extends http.BaseClient {
final http.Client _httpClient;
final Duration timeout;
TimeoutClient(
{http.Client? httpClient, this.timeout = const Duration(seconds: 30)})
: _httpClient = httpClient ?? http.Client();
Future<http.StreamedResponse> send(http.BaseRequest request) async {
// wait for result between two Futures (the one that is reached first) in silent mode (no throw exception)
_Either<http.StreamedResponse, Exception> result = await Future.any([
Future.delayed(
timeout,
() => _Either.Right(
TimeoutException(
'Client connection timeout after ${timeout.inMilliseconds} ms.'),
)),
Future(() async {
try {
return _Either.Left(await _httpClient.send(request));
} on Exception catch (e) {
return _Either.Right(e);
}
})
]);
// this code is reached only for first Future response,
// the second Future is ignorated and does not reach this point
if (result.right != null) {
throw result.right!;
}
return result.left!;
}
}
Their is onError option which works fine if their is any exception like no internet.It has to return response(my case in below code) or null.
In response their are 2 options Body and Status code.
var response = await http.post(url, body: body, headers: _headers).onError(
(error, stackTrace) => http.Response(
jsonEncode({
'message':no internet please connect to internet first
}),
408));
You can also choose to override the settings for a HttpClient:
class DevHttpOverrides extends HttpOverrides {
#override
HttpClient createHttpClient(SecurityContext? context) {
return super.createHttpClient(context)
..connectionTimeout = Duration(seconds: 2);
}
}
I am trying to make a simple request to backend using rxDart. But the problem I face is that when I get a http error such as 404, onError is not called, however, it is possible to extract it in onData.
I have a little experience with RxJava + retrofit and there it works as expected, when there is a response with error http status code onError is called and can be handled appropriately.
1. What am I doing wrong, or is it intended behavior?.
Object sendProfileData() {
Stream<Response> stream = onboardingRepository.createUser(User(name: 'name', surname: 'surname', lat: 1.0, lng: 2.0));
stream.listen((response) {
print(response.statusCode);
setAttributes();
}, onError: (e) {
print(e);
});
}
OnboardingRepository.dart:
class OnboardingRepository {
Observable<Response> createUser(User user) {
return Observable.fromFuture(TMApi.createUser(user));
}
}
TMApi.dart:
class TMApi {
static Future<http.Response> createUser(User user) async {
String url = '$baseUrl/create_user';
return await http.post(url, body: json.encode(user.toJson()));
}
}
What would be the best way to handle the event in the View? There should be an error displayed if error occurs, otherwise it should open a new screen. sendProfileData() method will return an Object, based on that I am going to perform actions in the view, but that doesn't sound like a very elegant solution...
Any suggestions on architecture are welcome :)
the http library in dart works a bit different than Retrofit.
The Future returned by http.post only throws an exception when there is an io error (socket error, no internet).
Server responses like 404 are reflected in the http.Response.
I created a simple convenience method that might help you:
void throwIfNoSuccess(http.Response response) {
if(response.statusCode < 200 || response.statusCode > 299) {
print('http error!');
print(response.body);
throw new HttpException(response);
}
}
class HttpException implements Exception {
HttpException(this.response);
http.Response response;
}
How to use:
import 'dart:convert';
import 'package:http/http.dart' as http;
Future<UserProfile> getUserProfile(String userId) async {
final url = 'https://example.com/api/users/$userId';
final response = await http.get(url);
throwIfNoSuccess(response);
final jsonBody = json.decode(response.body);
return UserProfile.fromJson(jsonBody);
}