Android Instrumentation Tests

Instrumentation tests are Java tests based on android.app.Instrumentation. They run on a device.

Writing an instrumentation test

Currently, an instrumentation test is just a JUnit3-style test based on android.test.InstrumentationTestCase. (This will change starting in Android N.)

Writing an instrumentation test case can be simple, e.g.

package org.chromium.sample.test;

import android.test.InstrumentationTestCase;

public class MyInstrumentationTest extends InstrumentationTestCase {

    // Note that, because it's a JUnit3-style test, the test method *must*
    // start with "test".
    public void testTheFirst() {
        bool writingInstrumentationTestsCanBeEasy = true;

        // InstrumentationTestCase inherits the assert* methods through
        // junit.framework.TestCase.
        assertTrue(writingInstrumentationTestsCanBeEasy);
    }

    public void testTheSecond() {
        bool writingInstrumentationTestsIsAlwaysEasy = false;
        assertFalse(writingInstrumentationTestsIsAlwaysEasy);
    }
}

After writing a test, you can run it by:

  • Adding the file to the relevant gn target if the entire file is new. Typically, the “relevant gn target” is simply the target containing the other files in the same directory.
  • Rebuild.
  • Run the test using the process described here.

Instrumentation test features

In many cases, Chromium has extended the instrumentation test framework classes to implement additional features.

Tracing

Enabling tracing during a test run allows all the function calls involved to be observed in a visual display (using Chrome's built-in chrome://tracing feature). To run a test with tracing, add the --trace-output flag to the command used to call the instrumentation test (either running the test_runner.py script, or a generated binary such as run_chrome_public_test_apk). The --trace-output flag takes a filename, which, after the test run, will contain a JSON file readable by chrome://tracing.

By default, the trace includes only certain function calls important to the test run, both within the Python test runner framework and the Java code running on the device. For a more detailed look, add the (no-argument) --trace-all flag. This causes every function called on the Python side to be added to the trace.

Annotations

Instrumentation tests in Chromium use a wide variety of annotations to control and manipulate test execution. Some of these are implemented in Chromium, while others are pulled in from outside. They include:

Test Batching

The @Batch("group_name") annotation is used to run all tests with the same batch group name in the same instrumentation invocation. In other words, the browser process is not restarted between these tests, and so any changes to global state, like launching an Activity, will persist between tests within a batch group. The benefit of this is that these tests run significantly faster - the per-test cost of restarting the process can be as high as 10 seconds (usually around 2 seconds), and that doesn't count the cost of starting an Activity like ChromeTabbedActivity.

Size annotations

Size annotations are used primarily by the test runner to determine the length of time to wait before considering a test hung (i.e., its timeout duration).

Several of the annotations are Android APIs from android.test.suitebuilder.annotation (prior to Android N) or android.support.test.filters (starting in Android N). These are all fairly self-explanatory:

A few additional size annotations are provided in //base:

  • @EnormousTest (timeout: 10 minutes) Typically used for tests that require WiFi.
  • @IntegrationTest (timeout: 30 minutes) Used for tests that run against real services.
  • @Manual (timeout: 10 hours) Used for manual tests.

Beware that the timeout durations for these annotations are subject to change, though they rarely do. These values are defined here.

Annotations that disable tests

There are several annotations that control whether or not a test runs. Some are conditional, others are not.

Unconditional disabling

@DisabledTest unconditionally disables a test.

@DisabledTest(
    // Describes why the test is disabled. Typically includes a crbug link.
    message = ""
)

@FlakyTest marks a test as flaky. This also unconditionally disables the test, though tests marked with @FlakyTest are explicitly run on some bots.

@FlakyTest(
    // Describes why the test is marked flaky. Typically includes a crbug link.
    message = ""
)

Note that there are Android versions of @DisabledTest and @FlakyTest that do not allow message specification. These are no longer used in Chromium.

As alluded to above, tests marked with either @DisabledTest or @FlakyTest can be explicitly run via the test runner's -A/--annotation flag. For example, this would run only the tests marked as flaky in chrome_public_test_apk:

./out/Debug/bin/run_chrome_public_test_apk -A FlakyTest
Conditional disabling

There are two primary annotation categories that conditionally disable tests: @DisableIf and @Restriction. The @DisableIf annotations are intended to temporarily disable a test in certain scenarios where it should work but doesn't. In contrast, the @Restriction annotation is intended to permanently limit a test to specific configurations. It signifies that the test was not, is not, and will not be intended to run beyond those configurations. In both cases, conditional disabling manifests as a skipped test.

@DisableIf.Build allows for conditional test disabling based on values in android.os.Build:

@DisableIf.Build(

    // Describes why the test is disabled.
    message = "",

    // Disables the test on SDK levels that match the given conditions.
    // Checks against Build.VERSION.SDK_INT.
    sdk_is_greater_than = 0,
    sdk_is_less_than = Integer.MAX_VALUE,

    // Disables the test on devices that support the given ABI
    // (e.g. "arm64-v8a"). Checks against:
    //  - Build.SUPPORTED_ABIS on L+
    //  - Build.CPU_ABI and Build.CPU_ABI2 otherwise
    supported_abis_includes = "",

    // Disables the test on devices with hardware that matches the given
    // value. Checks against Build.HARDWARE.
    hardware_is = "",

    // Disables the test on devices with product names that contain the
    // given value. Checks against Build.PRODUCT.
    product_name_includes = "",

)

@DisableIf.Device allows for conditional test disabling based on whether a device is a phone, a tablet, or a “large tablet” as determined by org.chromium.ui.base.DeviceFormFactor. This is available to tests in //ui or code that uses //ui.

@DisableIf.Device(
    // Disables the test on devices that match the given type(s) as described
    // above.
    type = {}
)

@Restriction currently allows for conditional test disabling based on device type, device performance, internet connectivity, whether Google Play Services is up to date, and whether the build was an official one.

@Restriction(
    // Possible values include:
    //
    // base:
    //  - Restriction.RESTRICTION_TYPE_LOW_END_DEVICE
    //    Restricts the test to low-end devices as determined by SysUtils.isLowEndDevice().
    //
    //  - Restriction.RESTRICTION_TYPE_NON_LOW_END_DEVICE
    //    Restricts the test to non-low-end devices as determined by SysUtils.isLowEndDevice().
    //
    //  - Restriction.RESTRICTION_TYPE_INTERNET
    //    Restricts the test to devices that have an internet connection.
    //
    // chrome:
    //  - ChromeRestriction.RESTRICTION_TYPE_GOOGLE_PLAY_SERVICES
    //    Restricts the test to devices with up-to-date versions of Google Play Services.
    //
    //  - ChromeRestriction.RESTRICTION_TYPE_OFFICIAL_BUILD
    //    Restricts the test to official builds as determined by ChromeVersionInfo.isOfficialBuild().
    //
    // ui:
    //  - UiRestriction.RESTRICTION_TYPE_PHONE
    //    Restricts the test to phones as determined by DeviceFormFactor.
    //
    //  - UiRestriction.RESTRICTION_TYPE_TABLET
    //    Restricts the test to tablets as determined by DeviceFormFactor.
    value = {}
)

@MinAndroidSdkLevel is similar to @Restriction in purpose in that it's intended to permanently limit a test to only recent versions of Android.

@MinAndroidSdkLevel(
    // The minimum SDK level at which this test should be executed. Checks
    // against Build.VERSION.SDK_INT.
    value = 0
)

Annotations that affect how a test is run

Several annotations affect how a test is run in interesting or nontrivial ways.

@CommandLineFlags.Add and @CommandLineFlags.Remove manipulate Chrome's command-line flags on a per-test basis (i.e., the flags handled by org.chromium.base.CommandLine and base::CommandLine).

@CommandLineFlags.Add(
    // The flags to add to the command line for this test. These can be
    // anything and typically should include the leading dashes (e.g. "--foo").
    value = {}
)

@CommandLineFlags.Remove(
    // The flags to remove from the command line for this test. These can only
    // be flags added via @CommandLineFlags.Add. Flags already present in the
    // command-line file on the device are only present in the native
    // CommandLine object and cannot be manipulated.
    value = {}
)

Feature annotations

@Feature has been used inconsistently in Chromium to group tests across test cases according to the feature they're testing.

@Feature(
    // The features associated with this test. These can be anything.
    value = {}
)

@Feature doesn‘t have an inherent function, but it can be used to filter tests via the test runner’s -A/--annotation and -E/--exclude-annotation flags. For example, this would run only the tests with @Feature annotations containing at least “Sync” in chrome_public_test_apk:

out/Debug/bin/run_chrome_public_test_apk -A Feature=Sync

Advanced