This module is part of the Apache Sling project.
Maven Plugin for building OSGi features and applications.
The format of a feature file is described in the Apache Sling Feature Model. As described there, a feature file must have an ID. If you use this Maven Plugin you have two options:
If you want to have full control over the ID, you can encode it directly in the feature file. However, it is advised to use placeholders for the group id, artifact id and version and let the maven plugin insert them for you.
In the following example the qualifier base is specified for the feature file:
{ "id":"${project.groupId}:${project.artifactId}:slingosgifeature:base:${project.version}", ... }
If you omit the ID in your feature file, the maven plugin will generate an ID for you based on the file name of the feature file. The file name becomes the qualifier of the ID. If your feature is named feature.json then this becomes the main artifact of the project without a qualifier.
This maven plugin replaces some placeholders when processing feature files, placeholders use the common ${name} syntax. The following placeholders are replaced by default:
The replacement of these placeholders can be disabled by setting the configuration enableProjectVariableReplacement to true.
The plugin can be configured to replace placeholders in the feature file based on maven project properties. To enable this, the configuration replacePropertyVariables can be set with a comma separated list of property names like:
<replacePropertyVariables>oak.version,jackrabbit.version</replacePropertyVariables>
The value needs to be set as a property on the maven project to be replaced.
Up to version 1.3.4 of the maven plugin, all occurences of placeholders where tried to be replaced by the maven plugin. However, this had two problems:
For versions higher than 1.3.4, the enableLegacyVariableReplacement configuration can be set to true to enable this old behaviour. But it is discouraged to do so and rather use replacePropertyVariables with a clearly defined set of placeholders.
For versions higher than 1.4.0, the maven plugin supports to author extensions in external files. For example a repoinit section can be written in a text file next to the feature model file. If an extension of type text is specified in the feature model with the value @file, the maven plugin looks for a file in the same directory as the feature model. The file must have the same name as the feature file followed by a dash, the name of the extension and “.txt”. For example a feature file src/main/features/myfeature.json with a repoinit extension using “@file” reads the repoinit from src/main/features/myfeature-repoinit.txt.
In addition, it is possible to control the name of the file name by specifying the part after the dash. For example using @file:xssapi.xml as the value of the extension leads to the a file named myfeature-xssapi.xml to be read.
For versions higher than 1.4.16, the maven plugin supports configuring global metadata for bundles and artifacts. For bundles as well for each extension with artifacts, a map of defaults can be configured. If the source of the feature does not have the configured property, it will be added. This allows to add for example the same metadata start order to bundles while single bundles might still override that value.
The below configuration sets the start-order to 5 and foo to bar for bundles, and start-order to 1 for content-packages.
<configuration> <defaultMetadata> <bundles> <!-- Defaults for bundles --> <start-order>5</start-order> <foo>bar</foo> </bundles> <content-packages> <!-- Defaults for an extension namend content-packages --> <start-order>1</start-order> </content-packages> </defaultMetadata> </configuration>
src/main/features.**/*.json, therefore all files with the extension .json are read including sub directories.This global configuration specifies the initial set of feature files used for the current project, the other goals can then refine this subset.
There might be other plugins involved which generate feature files are part of the build. This plugin by default looks into a directory named generated-features inside the build directory for such files. Including of generated files can be controlled with the following configurations:
**/*.json, therefore all files with the extension .json are read including sub directories.Most of the plugin goals take a selection of features as input, for example to aggregate a set of features to a new feature or to analyse a specific set of features according to some rules.
All of these goals use the same way of selecting the features: Whenever a goal can select from the above global list of features, filesInclude and filesExclude can be used to select from the above list of project files. The patterns are relative to the specified features directory:
... <!-- Include all feature files with a filename starting with base --> <filesInclude>**/base-*.json</filesInclude> <!-- Include a specific file --> <filesInclude>additional/special-feature.json</filesInclude> <!-- Exclude this file --> <filesExclude>connectors/base-http.json</filesExclude> ...
The order of the above include statements defines the order in which the features are processed. If an include contains a pattern, all files matching that pattern are processed in string order based on their full path.
In addition, most of the goals can also be configured to select aggregated features (see below) based on their qualifier. The special token * can be used to select all aggregated features and the token : selects the aggregated feature without a classifier (main artifact).
... <includeClassifier>core-aggregate</includeClassifier> <includeClassifier>web-aggregate</includeClassifier> ...
Again the order of the instructions defines the order of processing.
Most of the goals also support including of ref files. A ref file is a text file where each line in the text file contains a Maven url like mvn:groupId/artifactId/version. Lines starting with a hash can contain comments. The provided patterns are relative to the project directory.
... <!-- Include all ref files within a source directory --> <refsInclude>src/main/references/*.ref</refsInclude> <!-- Include ref files generated by another plugin --> <refsInclude>target/generated-refs/*.ref</refsInclude> ...
Finally, most of the goals also support to add features from other projects:
... <includeArtifact> <groupId>org.apache.sling</groupId> <artifactId>somefeature</artifactId> <version>1.0.0</version> <type>slingosgifeature</type> </includeArtifact> ...
All of the above ways to select features (project files, project aggregates and external features) can be mixed in the configuration. It's possible to first specify an artifact include, followed by a aggregate classifier, followed by file includes. And this is then the order of processing. Please note that file excludes can be placed anywhere and regardless of their position, they are always applied to every files include.
Produce an aggregated feature from a list of features. The list of features is either specified by include/exclude patterns based on the configured features directory of the project or Maven coordinates of features.
Sample configuration:
<plugin> <groupId>org.apache.sling</groupId> <artifactId>slingfeature-maven-plugin</artifactId> <extensions>true</extensions> <executions> <execution> <id>merge-features</id> <goals> <goal>aggregate-features</goal> </goals> <configuration> <aggregates> <!-- A list of feature aggregations, each aggregate creates a new feature: --> <aggregate> <classifier/> <!-- optional classifier or main artifact (no classifier)--> <title/> <!-- optional title--> <description/> <!-- optional description--> <vendor/> <!-- optional description--> <markAsFinal/> <!-- optional flag to mark the feature as final --> <markAsComplete/> <!-- optional flag to mark the feature as complete --> <filesInclude/> <!-- optional include for local files, this can be specified more than once --> <filesExclude/> <!-- optional exclude for local files, this can be specified more than once --> <includeArtifact/> <!-- optional artifact for external features, this can be specified more than once --> <includeClassifier/> <!-- optional classifier for aggregates, this can be specified more than once --> <variablesOverrides> <!-- Feature variables can be specified/overridden here --> <https.port>8443</https.port> <some.variable/> <!-- set some.variable to null --> </variablesOverrides> <frameworkPropertiesOverrides> <!-- Framework property overrides go here --> <org.osgi.framework.bootdelegation>sun.*,com.sun.*</org.osgi.framework.bootdelegation> </frameworkPropertiesOverrides> <artifactsOverrides> <!-- Artifact clash overrides go here --> <artifactsOverride>org.apache.sling:abundle:LATEST</artifactsOverride> <artifactsOverride>org.apache.sling:anotherbundle:1.4.5</artifactsOverride> </artifactsOverrides> </aggregate> </aggregates> </configuration> </execution> </executions> </plugin>
All features found in the directory as well as the artifact sections of the plugin configuration are aggregated into a single feature. Includes are processed in the way they appear in the configuration. If an include contains a pattern which includes more than one feature, than the features are included based on their full alphabetical file path. The features are aggregated in the order they are included.
If an include or an exclude is not using a pattern but directly specifying a file, this file must exists. Otherwise the build fails.
The merged feature will have the same groupId, artifactId and version as the pom in which the aggregation is configured. It will have type slingosgifeature and as classifier the one specified in the configuration named classifier.
Variables and framework properties can be overridden using the <variables> and <fraweworkProperties> sections. The values configured are only used if at least one feature contains such a variable or framework property. If multiple definitions of the same variable or framework properties are found in the features that are to be aggregated and the values for these variable / framework properties are different, they must be overridden, otherwise the aggregation will generate an error.
If the aggregation sources contain the same artifact more than once in different versions, a conflict resolution override must be specified using the <artifactsOverride> tag. In most cases the desired result will only contain one version of the artifact, although it's also possible to state that all versions must be kept.
The following syntax is supported for <artifactsOverride>:
<configuration> <aggregates> <aggregate> ... <!-- The overrides are processed in the specified order --> <artifactsOverrides> <!-- To provide a conflict resolution for a specific artifact: --> <artifactsOverride>groupid1:artifactid1:RESOLUTION</artifactsOverride> <!-- To apply the same override rule for all clashes, a wildcard using '*' for groupID and artifactID can be used; this can be used as a catch all --> <artifactsOverride>*:*:RESOLUTION</artifactsOverride> </artifactsOverrides> ...
The RESOLUTION is one of the following:
ALL - select all the artifacts, i.e. keep them side-by-side.HIGHEST - select only the artifact with the highest version number.LATEST - select only the artifact provided latest, so the version used in the last feature file merged in.FIRST - select only the artifact provided first, the version used in the first feature<version> - selects this specific version.When comparing version numbers these are converted to OSGi version numbers and the OSGi version number ordering is applied.
Merging of extensions is specific to the extension being merged. Handlers can be provided to implement the logic of extension merging. A handler needs to implement the org.apache.sling.feature.builder.FeatureExtensionHandler and is looked up via the Java ServiceLoader mechanism.
To provide additional handlers to the slingfeature-maven-plugin, list the artifacts in the <dependencies> section of the plugin configuration:
<plugin> <groupId>org.apache.sling</groupId> <artifactId>slingfeature-maven-plugin</artifactId> <executions> ... </executions> <dependencies> <dependency> <groupId>org.apache.sling</groupId> <artifactId>my-feature-extension-handler</artifactId> <version>1.0.0</version> </dependency> </dependencies> </plugin>
Run feature model analysers on the feature models in the project. Analysers are defined in the Apache Sling Feature Analyser project and are selected by their ID, which is obtained from the getId() method in Analyser Task.
<execution> <id>analyze</id> <goals> <goal>analyse-features</goal> </goals> <configuration> <scans> <!-- A list of scans, each creates a new analysis: --> <scan> <!-- specify which feature files to include --> <!-- optional include for local files, this can be specified more than once --> <filesInclude>**/*.json</filesInclude> <!-- optional exclude for local files, this can be specified more than once --> <filesExclude>dontcheck.json</filesExclude> <!-- optional classifier for aggregates, this can be specified more than once --> <includeClassifier>aggregated</includeClassifier <!-- optional artifact for external features, this can be specified more than once --> <includeArtifact> <groupId>org.apache.sling</groupId> <artifactId>somefeature</artifactId> <version>1.0.0</version> <type>slingosgifeature</type> </includeArtifact> <!-- if only a subset of tasks need to be run, specify them here --> <includeTask>api-regions-dependencies</includeTask> <!-- can also exclude tasks --> <excludeTask>do-not-run-this-task</excludeTask> <!-- taskConfiguration is a String, Properties map --> <taskConfiguration> <!-- the configuration is specific to the analyser tasks --> <!-- each key represents an AnalyserTask ID --> <api-regions-dependencies> <!-- can use keyname-textvalue syntax to specify configuration for a task --> <exporting-apis>global</exporting-apis> <hiding-apis>internal</hiding-apis> </api-regions-dependencies> </taskConfiguration> </scan> </scans> </configuration> </execution>
Attach feature files found in the project to the projects produced artifacts. This includes features found in src/main/features as well as features produce with the aggregate-features goal if no configuration is specified.
A feature archive contains one or more features together with all artifacts referenced by the feature like bundles or other binary artifacts. With this goal such an archive can be created for features and attached to the project.
By default all features known to the project are used to create feature archive for each feature, this includes features which are part of the source of the project, features generated by other plugins and aggregates.
It is possible to refine which feature archives should be created by selecting features as well as creating multiple features. The below configuration creates two feature archives. The first one with the classifier platformarchive contains a single feature with the classifier platform. The second archive gets the classifier uiarchive and contains two features with the classifiers uibase and ui.
By default, a feature archive gets the type far. By specifying the optional type configuration this can be changed to another type.
<execution> <id>create-feature-archives</id> <goals> <goal>attach-featurearchives</goal> </goals> <configuration> <archives> <archive> <classifier>platformarchive</classifier> <includeClassifier>platform</includeClassifier> // optional type <type>zip</type> </archive> <archive> <classifier>uiarchive</classifier> <includeClassifier>uibase</includeClassifier> <includeClassifier>ui</includeClassifier> </archive> </archives> </configuration> </execution>
This goal can be used to extract the contents of an extension into a local file, which may be useful for other tools that can work on the content of the extension.
The goal is configured as in the following example:
<execution> <id>extract-apiregions</id> <goals> <goal>extract-extension</goal> </goals> <configuration> <selection> <featuresInclude>feature-abc.json</featuresInclude> <extension>api-regions</extension> <outputDir>target/extracted</outputDir> </selection> </configuration> </execution>
This example extracts the api-region extension from the feature files feature-abc.json and puts it in the target/extracted directory. Feature files are selected as described in the Global Configuration above.
Output files are written to the output directory follows where the file name is the classifier of the selected feature, followed by the extension name and a .json extension for JSON and .txt extension for others.
Dependencies get out of date over time, with the update-feature-versions goal, all artifacts in a feature can be checked for available updates and updated to a newer version. By default all artifacts in all feature files in the project are checked and updated:
mvn slingfeature:update-feature-versions
To first check which artifacts will be updated and to what version, it is possible to just do a dry runs which does not alter the feature files:
mvn slingfeature:update-feature-versions -DdryRun=true
The selection of feature files can be further refined by specifying the classifiers parameter which selects feature files based on their classifier. The special token ‘:’ can be used to select the main artifact (artifact without a classifier). The parameter takes a comma separated list:
mvn slingfeature:update-feature-versions -Dclassifiers=platform,:
By default all artifacts in a feature file are checked for updates. By specifying includes and excludes based on Maven coordinates the selection of artifacts can be further refined. The most common use case is to specify the group id only. For example the following command just updates all artifacts with the Sling group id:
mvn slingfeature:update-feature-versions -Dincludes=org.apache.sling
Both parameters, includes and excludes, take a comma separated list.
It is also possible to define the version scope, which means the policy how to update. By default ANY is used, meaning the latest found version (excluding snapshots) is used. You can also specify MAJOR, MINOR, INCREMENTAL and SUBINCREMENTAL by using the versionScope parameter.
This can also be combined with includes to specify different scopes for each include:
mvn slingfeature:update-feature-versions -Dincludes=org.apache.sling/MAJOR,org.apache.felix/INCREMENTAL
Instead of specifying a scope, includes can also be used to define a specific version:
mvn slingfeature:update-feature-versions -Dincludes=org.apache.jackrabbit.oak/4.0.2
With the repository goal, a directory with all artifacts from the selected features will be created.
<execution> <id>repo</id> <goals> <goal>repository</goal> </goals> <configuration> <repositories> <!-- A list of repositories, each creates a new repository: --> <repository> <repositoryDir/> <!-- optional repository directory--> <filesInclude/> <!-- optional include for local files, this can be specified more than once --> <filesExclude/> <!-- optional exclude for local files, this can be specified more than once --> <includeArtifact/> <!-- optional artifact for external features, this can be specified more than once --> <includeClassifier/> <!-- optional classifier for aggregates, this can be specified more than once --> <embedArtifact/> <!-- optional artifact to be embedded in the repository. This can be specified more than once --> </repository> </repositories> </configuration> </execution>
The info goal allows to extract information about one or more features and generate some reports. It can be used within a Maven project or standalone.
The reports configuration should be configured with a comma separated list of the reports to be generated. The outputFormat configuration can be configured with either log in which case all output goes to the log or file (the default) in which case files for each feature and report are generated or singlefile in which case a single file for all report across all features is generated.
This report is done across all specified features and reports duplicates (bundles, configurations, artifacts or framework properties)
A text file with a list of exported packages is generated for each feature. If a feature does not export packages, no file is created.
This report is done across all features and list the contents (bundles and artifacts).
A text file with a list of imported packages is generated for each feature. If a feature does not import packages, no file is created.
When used in a Maven project, the list of features for the input goal can be specified by defining the feature set for the goal. The features are configured as described in the global configuration section above. If no configuration is provided, all feature files from the project are used.
<execution> <id>info</id> <goals> <goal>info</goal> </goals> <configuration> <infoFeatures> <!-- specify which feature files to include --> <!-- optional include for local files, this can be specified more than once --> <filesInclude>**/*.json</filesInclude> </infoFeatures> </configuration> </execution>
The info goal can also be used standalone without a Maven project. In this case the property infoFeatureFiles must be specified with a comma separated list of feature files:
mvn slingfeature:info -DinfoFeatureFiles=/path/to/my/feature.json
If the outputFormat is set to file, by default the files are generated in the current directory. outputDirectory can be used to specify an alternative directory.
To generate the duplicates report to the log use:
mvn slingfeature:info -Dreports=duplicates -DoutputFormat=log
If the outputFormat is set to file, by default the files are generated a directory named feature-reports in the build directory. outputDirectory can be used to specify an alternative directory.
Attention: This Mojo is BETA meaning under development and new released may change the way the Mojo is used.
This Mojo allows the user to launch a Feature(s) from a POM through a profile. This can look like this:
<profile> <id>launch</id> <build> <plugins> <plugin> <groupId>org.apache.sling</groupId> <artifactId>slingfeature-maven-plugin</artifactId> <extensions>true</extensions> <dependencies> <!-- To Support the Deployment of Content Package the Extension Content must be added BEFORE the Feature Launcher --> <dependency> <groupId>org.apache.sling</groupId> <artifactId>org.apache.sling.feature.extension.content</artifactId> <version>LATEST_VERSION</version> </dependency> <dependency> <groupId>org.apache.sling</groupId> <artifactId>org.apache.sling.feature.launcher</artifactId> <version>LATEST_VERSION</version> </dependency> </dependencies> <executions> <execution> <id>launch-it</id> <phase>install</phase> <goals> <goal>launch-features</goal> </goals> <configuration> <selection> <includeClassifier>example-runtime</includeClassifier> </selection> </configuration> </execution> </executions> </plugin> </plugins> </build> </profile>
Do avoid having to release this plugin every time a new Feature Launcher is released the two are decoupled and the Feature Launcher is loaded at runtime instead. For that the feature launcher must be added to the plugin as dependency. If missing this Mojo with fail as it cannot find the launcher.
Attention: to deploy converted Content Packages the Feature Content Extension must added here as well and it must be place AHEAD of the feature launcher.
Beside the Feature Files this Mojo for now supports all the parameters of the current Feature Launcher (1.0.7-SNAPSHOT). For more info see the FeautreLaucherMojoTest.testFullLaunch() test method.
Support for Feature Archives: feature archives (FAR) are supported now and can be added to the project either:
By its classifier if part of the project
<featureArchiveClassifiers> <featureArchiveClassifier>sling12archive</featureArchiveClassifier> </featureArchiveClassifiers>
Or by its Artifact Id when installed in the local Maven Repo
<featureArchiveIds> <featureArchiveId>org.apache.sling.sample.far:sling-sample-far:far:samplefararchive:1.0.0-SNAPSHOT</featureArchiveId> </featureArchiveIds>
With the include-artifact goal it is possible to generate a POM based Feature Model so that it later can be used to be incorporated into other Feature Models without having to reference or copy the Feature Model files.
This snippet is creating a POM based Feature Model (create-test) and then incorporates another Feature Model (feature-test-repoinit.json) which contains the Repository Init instruction for that Feature Model. Eventually the result is installed (attach-features) into the local Maven repository.
<!-- Generates and Installs the Sling OSGi Feature Model file --> <plugin> <groupId>org.apache.sling</groupId> <artifactId>slingfeature-maven-plugin</artifactId> <extensions>true</extensions> <executions> <execution> <id>create-fm</id> <phase>package</phase> <goals> <goal>include-artifact</goal> </goals> <configuration> <includeArtifactClassifier>create-test</includeArtifactClassifier> </configuration> </execution> <execution> <id>aggregate</id> <phase>package</phase> <goals> <goal>aggregate-features</goal> </goals> <configuration> <aggregates> <aggregate> <!-- This must reference the classifier fro the 'create-fm' execution above --> <includeClassifier>create-test</includeClassifier> <filesInclude>feature-test-repoinit.json</filesInclude> </aggregate> </aggregates> </configuration> </execution> <execution> <id>attach</id> <phase>package</phase> <goals> <goal>attach-features</goal> </goals> </execution> </executions>
The aggregate execution above is only necessary if additional Feature Model Snippets are added to the final Feature Model file. See the src/it/include-artifact folders for more details.
The Feature Model Descriptor file name in the local Maven repository can be found here:
<group id as path>/<artifact-id>/<version>/<artifact id>-<version>.slingosgifeature
This MOJO compares different versions of the same Feature Model, producing the prototype Feature Model that shows the differences.
<execution> <id>diff</id> <goals> <goal>features-diff</goal> </goals> <configuration> <selection> <includeClassifier>cloud-ready</includeClassifier> <includeClassifier>main-jar</includeClassifier> </selection> <comparisonVersion>1.0.0-beta</comparisonVersion> <!-- (,${project.version}) by default --> </configuration> </execution>
It will produce ${project.classifier}.json JSON output files under the ${project.build.directory}/features-diff directory, where IDs will be composed by:
${project.groupId}:${project.artifactId}:${project.classifier}_updater:${project.version}
Diff JSON file will be attached to the project.