Mon utilitaire répète l'exécution exécutable ou appelable jusqu'à ce qu'il réussisse sans erreur ou lance une exécution après un délai. Cela fonctionne parfaitement pour les tests Espresso!
Supposons que la dernière interaction de vue (clic sur le bouton) active certains threads d'arrière-plan (réseau, base de données, etc.). En conséquence, un nouvel écran devrait apparaître et nous voulons le vérifier à notre prochaine étape, mais nous ne savons pas quand le nouvel écran sera prêt à être testé.
L'approche recommandée consiste à forcer votre application à envoyer des messages sur les états des threads à votre test. Parfois, nous pouvons utiliser des mécanismes intégrés comme OkHttp3IdlingResource. Dans d'autres cas, vous devez insérer des morceaux de code à différents endroits des sources de votre application (vous devez connaître la logique de l'application!) Pour tester la prise en charge uniquement. De plus, nous devrions désactiver toutes vos animations (bien que ce soit la partie de l'interface utilisateur).
L'autre approche attend, par exemple SystemClock.sleep (10000). Mais nous ne savons pas combien de temps attendre et même de longs retards ne peuvent garantir le succès. En revanche, votre test durera longtemps.
Mon approche consiste à ajouter une condition de temps pour visualiser l'interaction. Par exemple, nous testons que le nouvel écran doit apparaître pendant 10000 mc (timeout). Mais nous n'attendons pas et le vérifions aussi vite que nous le souhaitons (par exemple toutes les 100 ms) Bien sûr, nous bloquons le fil de test de cette manière, mais généralement, c'est exactement ce dont nous avons besoin dans de tels cas.
Usage:
long timeout=10000;
long matchDelay=100; //(check every 100 ms)
EspressoExecutor myExecutor = new EspressoExecutor<ViewInteraction>(timeout, matchDelay);
ViewInteraction loginButton = onView(withId(R.id.login_btn));
loginButton.perform(click());
myExecutor.callForResult(()->onView(allOf(withId(R.id.title),isDisplayed())));
Voici ma source de classe:
/**
* Created by alexshr on 02.05.2017.
*/
package com.skb.goodsapp;
import android.os.SystemClock;
import android.util.Log;
import java.util.Date;
import java.util.concurrent.Callable;
/**
* The utility repeats runnable or callable executing until it pass without errors or throws throwable after timeout.
* It works perfectly for Espresso tests.
* <p>
* Suppose the last view interaction (button click) activates some background threads (network, database etc.).
* As the result new screen should appear and we want to check it in our next step,
* but we don't know when new screen will be ready to be tested.
* <p>
* Recommended approach is to force your app to send messages about threads states to your test.
* Sometimes we can use built-in mechanisms like OkHttp3IdlingResource.
* In other cases you should insert code pieces in different places of your app sources (you should known app logic!) for testing support only.
* Moreover, we should turn off all your animations (although it's the part on ui).
* <p>
* The other approach is waiting, e.g. SystemClock.sleep(10000). But we don't known how long to wait and even long delays can't guarantee success.
* On the other hand your test will last long.
* <p>
* My approach is to add time condition to view interaction. E.g. we test that new screen should appear during 10000 mc (timeout).
* But we don't wait and check new screen as quickly as it appears.
* Of course, we block test thread such way, but usually it's just what we need in such cases.
* <p>
* Usage:
* <p>
* long timeout=10000;
* long matchDelay=100; //(check every 100 ms)
* EspressoExecutor myExecutor = new EspressoExecutor<ViewInteraction>(timeout, matchDelay);
* <p>
* ViewInteraction loginButton = onView(withId(R.id.login_btn));
* loginButton.perform(click());
* <p>
* myExecutor.callForResult(()->onView(allOf(withId(R.id.title),isDisplayed())));
*/
public class EspressoExecutor<T> {
private static String LOG = EspressoExecutor.class.getSimpleName();
public static long REPEAT_DELAY_DEFAULT = 100;
public static long BEFORE_DELAY_DEFAULT = 0;
private long mRepeatDelay;//delay between attempts
private long mBeforeDelay;//to start attempts after this initial delay only
private long mTimeout;//timeout for view interaction
private T mResult;
/**
* @param timeout timeout for view interaction
* @param repeatDelay - delay between executing attempts
* @param beforeDelay - to start executing attempts after this delay only
*/
public EspressoExecutor(long timeout, long repeatDelay, long beforeDelay) {
mRepeatDelay = repeatDelay;
mBeforeDelay = beforeDelay;
mTimeout = timeout;
Log.d(LOG, "created timeout=" + timeout + " repeatDelay=" + repeatDelay + " beforeDelay=" + beforeDelay);
}
public EspressoExecutor(long timeout, long repeatDelay) {
this(timeout, repeatDelay, BEFORE_DELAY_DEFAULT);
}
public EspressoExecutor(long timeout) {
this(timeout, REPEAT_DELAY_DEFAULT);
}
/**
* call with result
*
* @param callable
* @return callable result
* or throws RuntimeException (test failure)
*/
public T call(Callable<T> callable) {
call(callable, null);
return mResult;
}
/**
* call without result
*
* @param runnable
* @return void
* or throws RuntimeException (test failure)
*/
public void call(Runnable runnable) {
call(runnable, null);
}
private void call(Object obj, Long initialTime) {
try {
if (initialTime == null) {
initialTime = new Date().getTime();
Log.d(LOG, "sleep delay= " + mBeforeDelay);
SystemClock.sleep(mBeforeDelay);
}
if (obj instanceof Callable) {
Log.d(LOG, "call callable");
mResult = ((Callable<T>) obj).call();
} else {
Log.d(LOG, "call runnable");
((Runnable) obj).run();
}
} catch (Throwable e) {
long remain = new Date().getTime() - initialTime;
Log.d(LOG, "remain time= " + remain);
if (remain > mTimeout) {
throw new RuntimeException(e);
} else {
Log.d(LOG, "sleep delay= " + mRepeatDelay);
SystemClock.sleep(mRepeatDelay);
call(obj, initialTime);
}
}
}
}
https://gist.github.com/alexshr/ca90212e49e74eb201fbc976255b47e0