Some test modules may require customized setup and tear down steps that cannot be performed within the test case itself. Typical examples may include:
- install other apks (in addition to the test apk)
- push some files to the device
- run commands (e.g. adb shell pm ...)
In the past, component teams usually resort to writing a host side test to perform such tasks, which requires understanding of Trade Federation harness and typically increases the complexity of a test module .
Borrowing from CTS, we introduced the concept of test module config to support such tasks, the common tasks list above can be achieved by just a few lines of config. For maximum flexibility, you can even implement your own target preparer, as defined by ITargetPreparer or ITargetCleaner, and configure them to use in your own test module config.
A test module config for a test module is a required XML file added to the top level module source folder, named ‘AndroidTest.xml’. The XML follows the format of a configuration file used by Trade Federation test automation harness. Currently the main tags handled via the test module configs are the “target_preparer” and "test" tags.
A “target_preparer” tag, as the name suggests, defines a target preparer (see ITargetPreparer) that offers a setup method, which gets called before the test module is executed for testing; and if the class referenced in the “target_preparer” tag also implements ITargetCleaner, its teardown method will be invoked after the test module has finished.
To use the built-in common module config, add a new file ‘AndroidTest.xml’ at the top level folder for your test module, and populate it with the following content:
<?xml version="1.0" encoding="utf-8"?> <!-- [insert standard AOSP copyright here] --> <configuration description="Test module config for Foo"> <!-- insert options here --> </configuration>
As an example, we can add the following option tags (at the “insert” comment above):
<target_preparer class="com.android.tradefed.targetprep.RunCommandTargetPreparer"> <option name="run-command" value="settings put secure accessibility_enabled 1" /> <option name="teardown-command" value="settings put secure accessibility_enabled 0" /> </target_preparer>
The options will configure the test harness to:
- before test module is invoked, execute shell command “settings put secure accessibility_enabled 1” on device
- after test module is finished, execute shell command “settings put secure accessibility_enabled 0”
In this particular example, accessibility is enabled/disabled before/after the test module execution, respectively. With a simple example demonstrated, it’s necessary to cover more details on how the “option” tag is used. As shown above, the tag can have two attributes: name, value. The name attribute indicated the name of the option, and is further broken down into two parts separated by a colon: short name for the preparer, and the actual option name offered by the preparer.
The exact purpose of value field is dependent on how preparer defined the option: it can be a string, a number, a boolean, or even a file path etc. In the example above, name “run-command:run-command” means that we are setting value for the option “run-command” defined by a target preparer with short name “run-command”; and name “run-command:teardown-command” means that we are setting value for the option “teardown-command” also defined by the same target preparer with short name “run-command”. Here's a summary of the three common target preparers:
class name: PushFilePreparer
- short name: push-file
- function: pushes arbitrary files under test case folder into destination on device
- this preparer can push from folder to folder, or file to file; that is, you cannot push a file under a folder on device: you must specify the destination filename under that folder as well
- push-file: A push-spec, specifying the local file to the path where it should be pushed on device. May be repeated. If multiple files are configured to be pushed to the same remote path, the latest one will be pushed.
- push: (deprecated) A push-spec, formatted as
/path/to/srcfile.txt->/path/to/destfile.txt' or '
/path/to/srcfile.txt->/path/to/destdir/'. May be repeated. This path may be relative to the test module directory or the out directory itself.
- post-push: A command to run on the device (with `
adb shell <your command>`) after all pushes have been attempted. Typical use case would be using chmod for permissions
class name: InstallApkSetup
- short name:install-apk
- function: pushes arbitrary apk files under into destination on device
- test-file-name: the name of the apk to be installed on to device.
- install-arg: Additional arguments to be passed to the pm install command, including leading dash, e.g. “-d". May be repeated
class name: RunCommandTargetPreparer
- short name: run-command
- function: executes arbitrary shell commands before or after test module execution
- run-command:adb shell command to run. May be repeated
- teardown-command:adb shell command to run during teardown phase. May be repeated
A test class is the Trade Federation class to use to execute the test.
<test class="com.android.tradefed.testtype.AndroidJUnitTest"> <option name="package" value="android.test.example.helloworld"/> <option name="runner" value="android.support.test.runner.AndroidJUnitRunner"/> </test>
Here are three common test classes:
class name: GTest
- short name: gtest
- function: A Test that runs a native test package on given device.
- native-test-device-path:The path on the device where native tests are located.
class name: InstrumentationTest
- short name: instrumentation
- function: A Test that runs an instrumentation test package on given device
- package:The manifest package name of the Android test application to run.
- class:The test class name to run.
- method:The test method name to run.
class name: AndroidJUnitTest
- function: A Test that runs an instrumentation test package on given device using the android.support.test.runner.AndroidJUnitRunner This is the main way to execute an instrumentation test.