This plugin downloads/installs Node and NPM locally for your project, runs npm install
, and then any combination of
Bower, Grunt, Gulp, Jspm,
Karma, or Webpack.
It's supposed to work on Windows, OS X and Linux.
If you prefer Yarn over NPM for your node package fetching,
this plugin can also download Node and Yarn and then run yarn install
for your project.
- Let you keep your frontend and backend builds as separate as possible, by reducing the amount of interaction between them to the bare minimum; using only 1 plugin.
- Let you use Node.js and its libraries in your build process without installing Node/NPM globally for your build system
- Let you ensure that the version of Node and NPM being run is the same in every build environment
- Not meant to replace the developer version of Node - frontend developers will still install Node on their laptops, but backend developers can run a clean build without even installing Node on their computer.
- Not meant to install Node for production uses. The Node usage is intended as part of a frontend build, running common javascript tasks such as minification, obfuscation, compression, packaging, testing etc.
Notice: This plugin does not support already installed Node or npm versions. Use the exec-maven-plugin
instead.
- Maven 3 and Java 1.8
- For Maven 2 support take a look at the wiki.
Include the plugin as a dependency in your Maven project. Change LATEST_VERSION
to the latest tagged version.
<plugins>
<plugin>
<groupId>com.github.eirslett</groupId>
<artifactId>frontend-maven-plugin</artifactId>
<!-- Use the latest released version:
https://repo1.maven.org/maven2/com/github/eirslett/frontend-maven-plugin/ -->
<version>LATEST_VERSION</version>
...
</plugin>
...
Have a look at the example project, to see how it should be set up: https://github.com/eirslett/frontend-maven-plugin/blob/master/frontend-maven-plugin/src/it/example project/pom.xml
- Installing node and npm
- Installing node and yarn
- Running
- Configuration
Recommendation: Try to run all your tasks via npm scripts instead of running bower, grunt, gulp etc. directly.
The versions of Node and npm are downloaded from https://nodejs.org/dist, extracted and put into a node
folder created
in your installation directory . Node/npm will only be "installed" locally to your project.
It will not be installed globally on the whole system (and it will not interfere with any Node/npm installations already
present).
<plugin>
...
<executions>
<execution>
<!-- optional: you don't really need execution ids, but it looks nice in your build log. -->
<id>install node and npm</id>
<goals>
<goal>install-node-and-npm</goal>
</goals>
<!-- optional: default phase is "generate-resources" -->
<phase>generate-resources</phase>
</execution>
</executions>
<configuration>
<nodeVersion>v4.6.0</nodeVersion>
<!-- optional: with node version greater than 4.0.0 will use npm provided by node distribution -->
<npmVersion>2.15.9</npmVersion>
<!-- optional: where to download node and npm from. Defaults to https://nodejs.org/dist/ -->
<downloadRoot>http://myproxy.example.org/nodejs/</downloadRoot>
</configuration>
</plugin>
You can also specify separate download roots for npm and node as they are stored in separate repos.
<plugin>
...
<configuration>
<!-- optional: where to download node from. Defaults to https://nodejs.org/dist/ -->
<nodeDownloadRoot>http://myproxy.example.org/nodejs/</nodeDownloadRoot>
<!-- optional: where to download npm from. Defaults to https://registry.npmjs.org/npm/-/ -->
<npmDownloadRoot>https://myproxy.example.org/npm/</npmDownloadRoot>
</configuration>
</plugin>
You can use Nexus repository Manager to proxy npm registries. See https://help.sonatype.com/display/NXRM3/Npm Registry
Notice: Remember to gitignore the node
folder, unless you actually want to commit it.
Instead of using Node with npm you can alternatively choose to install Node with Yarn as the package manager.
The versions of Node and Yarn are downloaded from https://nodejs.org/dist
for Node
and from the Github releases for Yarn,
extracted and put into a node
folder created in your installation directory.
Node/Yarn will only be "installed" locally to your project.
It will not be installed globally on the whole system (and it will not interfere with any Node/Yarn installations already
present).
Have a look at the example POM
to see how it should be set up with Yarn:
https://github.com/eirslett/frontend-maven-plugin/blob/master/frontend-maven-plugin/src/it/yarn-integration/pom.xml
<plugin>
...
<execution>
<!-- optional: you don't really need execution ids, but it looks nice in your build log. -->
<id>install node and yarn</id>
<goals>
<goal>install-node-and-yarn</goal>
</goals>
<!-- optional: default phase is "generate-resources" -->
<phase>generate-resources</phase>
</execution>
<configuration>
<nodeVersion>v6.9.1</nodeVersion>
<yarnVersion>v0.16.1</yarnVersion>
<!-- optional: where to download node from. Defaults to https://nodejs.org/dist/ -->
<nodeDownloadRoot>http://myproxy.example.org/nodejs/</nodeDownloadRoot>
<!-- optional: where to download yarn from. Defaults to https://github.com/yarnpkg/yarn/releases/download/ -->
<yarnDownloadRoot>http://myproxy.example.org/yarn/</yarnDownloadRoot>
</configuration>
</plugin>
All node packaged modules will be installed in the node_modules
folder in your working directory.
By default, colors will be shown in the log.
<execution>
<id>npm install</id>
<goals>
<goal>npm</goal>
</goals>
<!-- optional: default phase is "generate-resources" -->
<phase>generate-resources</phase>
<configuration>
<!-- optional: The default argument is actually
"install", so unless you need to run some other npm command,
you can remove this whole <configuration> section.
-->
<arguments>install</arguments>
</configuration>
</execution>
Notice: Remember to gitignore the node_modules
folder, unless you actually want to commit it. Npm packages will
always be installed in node_modules
next to your package.json
, which is default npm behavior.
As with npm above, all node packaged modules will be installed in the node_modules
folder in your working directory.
<execution>
<id>yarn install</id>
<goals>
<goal>yarn</goal>
</goals>
<configuration>
<!-- optional: The default argument is actually
"install", so unless you need to run some other yarn command,
you can remove this whole <configuration> section.
-->
<arguments>install</arguments>
</configuration>
</execution>
All bower dependencies will be installed in the bower_components
folder in your working directory.
<execution>
<id>bower install</id>
<goals>
<goal>bower</goal>
</goals>
<configuration>
<!-- optional: The default argument is actually
"install", so unless you need to run some other bower command,
you can remove this whole <configuration> section.
-->
<arguments>install</arguments>
</configuration>
</execution>
Notice: Remember to gitignore the bower_components
folder, unless you actually want to commit it.
It will run Grunt according to the Gruntfile.js
in your working directory.
By default, colors will be shown in the log.
<execution>
<id>grunt build</id>
<goals>
<goal>grunt</goal>
</goals>
<!-- optional: the default phase is "generate-resources" -->
<phase>generate-resources</phase>
<configuration>
<!-- optional: if not specified, it will run Grunt's default
task (and you can remove this whole <configuration> section.) -->
<arguments>build</arguments>
</configuration>
</execution>
Very similar to the Grunt execution. It will run gulp according to the gulpfile.js
in your working directory.
By default, colors will be shown in the log.
<execution>
<id>gulp build</id>
<goals>
<goal>gulp</goal>
</goals>
<!-- optional: the default phase is "generate-resources" -->
<phase>generate-resources</phase>
<configuration>
<!-- optional: if not specified, it will run gulp's default
task (and you can remove this whole <configuration> section.) -->
<arguments>build</arguments>
</configuration>
</execution>
All jspm dependencies will be installed in the jspm_packages
folder in your working directory.
<execution>
<id>jspm install</id>
<goals>
<goal>jspm</goal>
</goals>
<configuration>
<!-- optional: The default argument is actually
"install", so unless you need to run some other jspm command,
you can remove this whole <configuration> section.
-->
<arguments>install</arguments>
</configuration>
</execution>
<execution>
<id>javascript tests</id>
<goals>
<goal>karma</goal>
</goals>
<!-- optional: the default plase is "test". Some developers
choose to run karma in the "integration-test" phase. -->
<phase>test</phase>
<configuration>
<!-- optional: the default is "karma.conf.js" in your working directory -->
<karmaConfPath>src/test/javascript/karma.conf.ci.js</karmaConfPath>
</configuration>
</execution>
Skipping tests: If you run maven with the -DskipTests
flag, karma tests will be skipped.
Ignoring failed tests: If you want to ignore test failures run maven with the -Dmaven.test.failure.ignore
flag,
karma test results will not stop the build but test results will remain
in test output files. Suitable for continuous integration tool builds.
Why karma.conf.ci.js? When using Karma, you should have two separate
configurations: karma.conf.js
and karma.conf.ci.js
. (The second one should inherit configuration
from the first one, and override some options. The example project shows you how to set it up.)
The idea is that you use karma.conf.js
while developing (using watch/livereload etc.), and
karma.conf.ci.js
when building - for example, when building, it should only run karma once,
it should generate xml reports, it should run only in PhantomJS, and/or it should generate
code coverage reports.
Running Karma through Grunt or gulp: You may choose to run Karma directly through Grunt
or through gulp instead, as part of the grunt
or gulp
execution. That
will help to separate your frontend and backend builds even more.
<execution>
<id>webpack build</id>
<goals>
<goal>webpack</goal>
</goals>
<!-- optional: the default phase is "generate-resources" -->
<phase>generate-resources</phase>
<configuration>
<!-- optional: if not specified, it will run webpack's default
build (and you can remove this whole <configuration> section.) -->
<arguments>-p</arguments>
</configuration>
</execution>
The working directory is where you've put package.json
and your frontend configuration files (Gruntfile.js
or
gulpfile.js
etc). The default working directory is your project's base directory (the same directory as your pom.xml
).
You can change the working directory if you want:
<plugin>
<groupId>com.github.eirslett</groupId>
<artifactId>frontend-maven-plugin</artifactId>
<!-- optional -->
<configuration>
<workingDirectory>src/main/frontend</workingDirectory>
</configuration>
</plugin>
Notice: Npm packages will always be installed in node_modules
next to your package.json
, which is default npm behavior.
The installation directory is the folder where your node and npm are installed. You can set this property on the different goals. Or choose to set it for all the goals, in the maven configuration.
<plugin>
<groupId>com.github.eirslett</groupId>
<artifactId>frontend-maven-plugin</artifactId>
<!-- optional -->
<configuration>
<installDirectory>target</installDirectory>
</configuration>
</plugin>
If you have configured proxy settings for Maven in your settings.xml file, the plugin will automatically use the proxy for downloading node and npm, as well as passing the proxy to npm commands.
Non Proxy Hosts: npm does not currently support non proxy hosts - if you are using a proxy and npm install is not downloading from your repository, it may be because it cannot be accessed through your proxy. If that is the case, you can stop the npm execution from inheriting the Maven proxy settings like this:
<configuration>
<npmInheritsProxyConfigFromMaven>false</npmInheritsProxyConfigFromMaven>
</configuration>
If you have configured proxy settings for Maven in your settings.xml file, the plugin will automatically pass the proxy to bower commands. If that is the case, you can stop the bower execution from inheriting the Maven proxy settings like this:
<configuration>
<bowerInheritsProxyConfigFromMaven>false</bowerInheritsProxyConfigFromMaven>
</configuration>
If you want to disable proxy for Yarn you can use yarnInheritsProxyConfigFromMaven
. When you have proxy settings in your settings.xml file if you don't use this param it will run code below with proxy settings, in some cases you don't want that. Adding this param into the configuration section will solve this issue
<execution>
<id>tests</id>
<goals>
<goal>yarn</goal>
</goals>
<phase>compile</phase>
<configuration>
<yarnInheritsProxyConfigFromMaven>false</yarnInheritsProxyConfigFromMaven>
<arguments>run test</arguments>
</configuration>
</execution>
If you need to pass some variable to Node, you can set that using the property environmentVariables
in configuration
tag of an execution like this:
<configuration>
<environmentVariables>
<!-- Simple var -->
<Jon>Snow</Jon>
<Tyrion>Lannister</Tyrion>
<!-- Var value take from maven properties -->
<NODE_ENV>${NODE_ENV}</NODE_ENV>
</environmentVariables>
</configuration>
Ignoring failed tests: If you want to ignore test failures in specific execution you can set that using the property maven.test.failure.ignore
in configuration tag of an execution like this:
<configuration>
<maven.test.failure.ignore>true</maven.test.failure.ignore>
</configuration>
Each frontend build tool and package manager allows skipping execution. This is useful for projects that contain multiple builds (such as a module containing Java and frontend code).
Note that if the package manager (npm or yarn) is skipped, other build tools will also need to be skipped because they would not have been downloaded. For example, in a project using npm and gulp, if npm is skipped, gulp must also be skipped or the build will fail.
Tools and property to enable skipping
- npm
-Dskip.npm
- yarn
-Dskip.yarn
- bower
-Dskip.bower
- grunt
-Dskip.grunt
- gulp
-Dskip.gulp
- jspm
-Dskip.jspm
- karma
-Dskip.karma
- webpack
-Dskip.webpack
This plugin contains support for M2E, including lifecycle mappings and support for incremental builds in Eclipse.
The install-node-and-npm
goal will only run on a full project build. The other goals support incremental builds
to avoid doing unnecessary work. During an incremental build the npm
goal will only run if the package.json
file
has been changed. The grunt
and gulp
goals have new srcdir
and triggerfiles
optional configuration options; if
these are set they check for changes in your source files before being run. See the wiki for more information.
During development, it's convenient to have the "npm", "bower", "grunt", "gulp" and "karma" commands available on the command line. If you want that, use those helper scripts!
Run $ mvn clean install
Please post any issues on the Github's Issue tracker. Pull requests are welcome! You can find a full list of contributors here.