JUnit

Description

This task runs tests from the JUnit testing framework. The latest version of the framework can be found at http://www.junit.org. This task has been tested with JUnit 3.0 up to JUnit 3.8.2; it won't work with versions prior to JUnit 3.0. It also works with JUnit 4.0, including "pure" JUnit 4 tests using only annotations and no JUnit4TestAdapter.

Note: This task depends on external libraries not included in the Apache Ant distribution. See Library Dependencies for more information.

Note: You must have junit.jar available. You can do one of:

1. Put both junit.jar and ant-junit.jar in ANT_HOME/lib.
2. Do not put either in ANT_HOME/lib, and instead include their locations in your CLASSPATH environment variable.
3. Add both JARs to your classpath using -lib.
4. Specify the locations of both JARs using a <classpath> element in a <taskdef> in the build file.
5. Leave ant-junit.jar in its default location in ANT_HOME/lib but include junit.jar in the <classpath> passed to <junit>. (since Ant 1.7)

See the FAQ for details.

Tests are defined by nested test or batchtest tags (see nested elements).

Parameters

By using the errorproperty and failureproperty attributes, it is possible to perform setup work (such as starting an external server), execute the test, clean up, and still fail the build in the event of a failure.

The filtertrace attribute condenses error and failure stack traces before reporting them. It works with both the plain and XML formatters. It filters out any lines that begin with the following string patterns:

   "junit.framework.TestCase"
   "junit.framework.TestResult"
   "junit.framework.TestSuite"
   "junit.framework.Assert."
   "junit.swingui.TestRunner"
   "junit.awtui.TestRunner"
   "junit.textui.TestRunner"
   "java.lang.reflect.Method.invoke("
   "sun.reflect."
   "org.apache.tools.ant."
   "org.junit."
   "junit.framework.JUnit4TestAdapter"
   " more"

Nested Elements

The <junit> task supports a nested <classpath> element that represents a PATH like structure.

As of Ant 1.7, this classpath may be used to refer to junit.jar as well as your tests and the tested code.

jvmarg

If fork is enabled, additional parameters may be passed to the new VM via nested <jvmarg> elements. For example:

<junit fork="yes">
  <jvmarg value="-Djava.compiler=NONE"/>
  ...
</junit>

would run the test in a VM without JIT.

<jvmarg> allows all attributes described in Command-line Arguments.

sysproperty

Use nested <sysproperty> elements to specify system properties required by the class. These properties will be made available to the VM during the execution of the test (either ANT's VM or the forked VM, if fork is enabled). The attributes for this element are the same as for environment variables.

<junit fork="no">
  <sysproperty key="basedir" value="${basedir}"/>
  ...
</junit>

would run the test in ANT's VM and make the basedir property available to the test.

syspropertyset

You can specify a set of properties to be used as system properties with syspropertysets.

since Ant 1.6.

env

It is possible to specify environment variables to pass to the forked VM via nested <env> elements. For a description of the <env> element's attributes, see the description in the exec task.

Settings will be ignored if fork is disabled.

bootclasspath

The location of bootstrap class files can be specified using this PATH like structure - will be ignored if fork is not true or the target VM doesn't support it (i.e. Java 1.1).

since Ant 1.6.

permissions

Security permissions can be revoked and granted during the execution of the class via a nested permissions element. For more information please see permissions

Settings will be ignored if fork is enabled.

since Ant 1.6.

assertions

You can control enablement of Java 1.4 assertions with an <assertions> subelement.

Assertion statements are currently ignored in non-forked mode.

since Ant 1.6.

modulepath

The location of modules can be specified using this PATH like structure.
The modulepath requires fork to be set to true.

since Ant 1.9.8

upgrademodulepath

The location of modules that replace upgradeable modules in the runtime image can be specified using this PATH like structure.
The upgrademodulepath requires fork to be set to true.

since Ant 1.9.8

formatter

The results of the tests can be printed in different formats. Output will always be sent to a file, unless you set the usefile attribute to false. The name of the file is determined by the name of the test and can be set by the outfile attribute of <test>.

There are four predefined formatters - one prints the test results in XML format, the other emits plain text. The formatter named brief will only print detailed information for testcases that failed, while plain gives a little statistics line for all test cases. Custom formatters that need to implement org.apache.tools.ant.taskdefs.optional.junit.JUnitResultFormatter can be specified.

If you use the XML formatter, it may not include the same output that your tests have written as some characters are illegal in XML documents and will be dropped.

The fourth formatter named failure (since Ant 1.8.0) collects all failing testXXX() methods and creates a new TestCase which delegates only these failing methods. The name and the location can be specified via Java System property or Ant property ant.junit.failureCollector. The value has to point to the directory and the name of the resulting class (without suffix). It defaults to java-tmp-dir/FailedTests.

test

Defines a single test class.

Tests can define their own formatters via nested <formatter> elements.

batchtest

Define a number of tests based on pattern matching.

batchtest collects the included resources from any number of nested Resource Collections. It then generates a test class name for each resource that ends in .java or .class.

Any type of Resource Collection is supported as a nested element, prior to Ant 1.7 only <fileset> has been supported.

Batchtests can define their own formatters via nested <formatter> elements.

Forked tests and tearDown

If a forked test runs into a timeout, Ant will terminate the Java VM process it has created, which probably means the test's tearDown method will never be called. The same is true if the forked VM crashes for some other reason.

Starting with Ant 1.8.0, a special formatter is distributed with Ant that tries to load the testcase that was in the forked VM and invoke that class' tearDown method. This formatter has the following limitations:

• It runs in the same Java VM as Ant itself, this is a different Java VM than the one that was executing the test and it may see a different classloader (and thus may be unable to load the test class).
• It cannot determine which test was run when the timeout/crash occurred if the forked VM was running multiple test. I.e. the formatter cannot work with any forkMode other than perTest and it won't do anything if the test class contains a suite() method.

If the formatter recognizes an incompatible forkMode or a suite method or fails to load the test class it will silently do nothing.

The formatter doesn't have any effect on tests that were not forked or didn't cause timeouts or VM crashes.

To enable the formatter, add a formatter like

<formatter classname="org.apache.tools.ant.taskdefs.optional.junit.TearDownOnVmCrash"
           usefile="false"/>

to your junit task.

ant.junit.enabletestlistenerevents magic property

Since Ant 1.8.2 the enableTestListenerEvents attribute of the task controls whether fine grained logging messages will be sent to the task's verbose log. In addition to this attribute Ant will consult the property ant.junit.enabletestlistenerevents and the value of the property overrides the setting of the attribute.

This property exists so that containers running Ant that depend on the additional logging events can ensure they will be generated even if the build file disables them.

Examples

<junit>
  <test name="my.test.TestCase"/>
</junit>

Runs the test defined in my.test.TestCase in the same VM. No output will be generated unless the test fails.

<junit printsummary="yes" fork="yes" haltonfailure="yes">
  <formatter type="plain"/>
  <test name="my.test.TestCase"/>
</junit>

Runs the test defined in my.test.TestCase in a separate VM. At the end of the test, a one-line summary will be printed. A detailed report of the test can be found in TEST-my.test.TestCase.txt. The build process will be stopped if the test fails.

<junit printsummary="yes" haltonfailure="yes">
  <classpath>
    <pathelement location="${build.tests}"/>
    <pathelement path="${java.class.path}"/>
  </classpath>

  <formatter type="plain"/>

  <test name="my.test.TestCase" haltonfailure="no" outfile="result">
    <formatter type="xml"/>
  </test>

  <batchtest fork="yes" todir="${reports.tests}">
    <fileset dir="${src.tests}">
      <include name="**/*Test*.java"/>
      <exclude name="**/AllTests.java"/>
    </fileset>
  </batchtest>
</junit>

Runs my.test.TestCase in the same VM, ignoring the given CLASSPATH; only a warning is printed if this test fails. In addition to the plain text test results, for this test a XML result will be output to result.xml. Then, for each matching file in the directory defined for ${src.tests} a test is run in a separate VM. If a test fails, the build process is aborted. Results are collected in files named TEST-name.txt and written to ${reports.tests}.

<target name="test">
    <property name="collector.dir" value="${build.dir}/failingTests"/>
    <property name="collector.class" value="FailedTests"/>
    <!-- Delete 'old' collector classes -->
    <delete>
        <fileset dir="${collector.dir}" includes="${collector.class}*.class"/>
    </delete>
    <!-- compile the FailedTests class if present --> 
    <javac srcdir="${collector.dir}" destdir="${collector.dir}"/>
    <available file="${collector.dir}/${collector.class}.class" property="hasFailingTests"/>
    <junit haltonerror="false" haltonfailure="false">
        <sysproperty key="ant.junit.failureCollector" value="${collector.dir}/${collector.class}"/>
        <classpath>
            <pathelement location="${collector.dir}"/>
        </classpath>
        <batchtest todir="${collector.dir}" unless="hasFailingTests">
            <fileset dir="${collector.dir}" includes="**/*.java" excludes="**/${collector.class}.*"/>
            <!-- for initial creation of the FailingTests.java -->
            <formatter type="failure"/>
            <!-- I want to see something ... -->
            <formatter type="plain" usefile="false"/>
        </batchtest>
        <test name="FailedTests" if="hasFailingTests">
            <!-- update the FailingTests.java -->
            <formatter type="failure"/>
            <!-- again, I want to see something -->
            <formatter type="plain" usefile="false"/>
        </test>
    </junit>
</target>

On the first run all tests are collected via the <batchtest/> element. Its plain formatter shows the output on the console. The failure formatter creates a java source file in ${build.dir}/failingTests/FailedTests.java which extends junit.framework.TestCase and returns from a suite() method a test suite for the failing tests.
On a second run the collector class exists and instead of the <batchtest/> the single <test/> will run. So only the failing test cases are re-run. The two nested formatters are for displaying (for the user) and for updating the collector class.

    <junit fork="true"
        jvm="${platform.java}">
        <jvmarg line="--patch-module ${module.name}=${build.test.classes}"/>
        <jvmarg line="--add-modules ${module.name}"/>
        <jvmarg line="--add-reads ${module.name}=ALL-UNNAMED"/>
        <jvmarg line="--add-exports ${module.name}/my.test=ALL-UNNAMED"/>
        <classpath>
            <pathelement path="${libs.junit}"/>
        </classpath>
        <modulepath>
            <pathelement path="${modules}:${build.classes}"/>
        </modulepath>
        <formatter type="plain"/>
        <test name="my.test.TestCase"/>
    </junit>

Runs my.test.TestCase as a white-box test in the forked VM given by the platform.java property. The junit library is a part of an unnamed module while the tested project and required modules are on the module path. The tests do not have module-info file and are executed in the project module given by module.name property.
The --patch-module java option executes the tests built into ${build.test.classes} in a module given by module.name property.
The --add-modules java option enables the tested module.
The --add-reads java option makes the unnamed module containing the junit readable by tested module.
The --add-exports java option makes the non-exported test package my.test accessible from the unnamed module containing the junit.

    <junit fork="true"
        jvm="${platform.java}">
        <jvmarg line="--add-modules ${test.module.name}"/>
        <jvmarg line="--add-exports ${test.module.name}/my.test=junit,ALL-UNNAMED"/>
        <modulepath>
            <pathelement path="${modules}:${build.classes}:${libs.junit}"/>
        </modulepath>
        <formatter type="plain"/>
        <test name="my.test.TestCase"/>
    </junit>

Runs my.test.TestCase as a black-box test in the forked VM given by the platform.java property. The junit library is used as an automatic module. The tests module-info requires the tested module and junit.
The --add-modules java option enables the test module.
The --add-exports java option makes the non-exported test package my.test accessible from the junit module and Ant's test runner. Another possibility is to export the test package in the tests module-info by exports my.test directive.