jbang - Unleash the power of Java

Intro

Want to learn, explore or use Java instantly without setup ?

Do you like Java but use python, groovy, kotlin or similar languages for scripts, experimentation and exploration ?

Ever wanted to just be able to run java from anywhere without any or very minimal setup ? Ever tried out Java 11 support for running .java files directly in your shell but felt it was a bit too cumbersome ?

Then try jbang which lets you do this:

$ jbang init --template=cli hello.java
$ jbang hello.java Max!
[jbang] Resolving dependencies...
[jbang]     Resolving info.picocli:picocli:4.5.0...Done
[jbang] Dependencies resolved
[jbang] Building jar...
Hello Max!
$ jbang hello.java -h
Usage: hello [-hV] <greeting>
hello made with jbang
      <greeting>   The greeting to print
  -h, --help       Show this help message and exit.
  -V, --version    Print version information and exit.

Instant cli app generated built using java and picocli as a dependency that was fetched as needed for the compilation and execution.

JBang goes beyond more than just easy scripting; you can use jbang to launch any kind of java application or library packaged as a jar available locally, via http/https download or in a Maven repository.

Table of Contents

Features

.java Scripting for Java 8 and upwards
.jsh via JShell from Java 9 and upwards
Works on [windows] Windows, [apple] OSX and [linux] Linux
Install using curl, power shell, SDKMan ([apple]/[linux]), Homebrew ([apple]), Chocolatey ([windows]) or Scoop ([windows])
If needed will automatically install Java and even a Java editor (vscodium) for editing
Installation of scripts to user PATH
Include multiple files and sources
Dependency declarations using //DEPS <gav> for automatic dependency resolution
Control compile and runtime options with //JAVAC_OPTIONS <flags> and //JAVA_OPTIONS <flags>
Compiled jar and Dependency resolution caching
native-image generation (--native)
Launch with debug enabled for instant debugging from your favorite IDE
Transparent launch of JavaFX Applications on Java 8 and higher
Can be used for writing plugins to other cli’s like kubectl
Init templates to get started easily (jbang init -t cli hello.java)
Generate gradle and IDE config with dependencies for easy editing in your favorite IDE (jbang edit myfile.java)

To use it install jbang and run jbang yourscript.java

Requirements

Tested and verified to use on OSX, Linux and Windows (incl. command.exe, cygwin and mingw shells).

Installation

To use jbang Java 8 is the minimum required version, however Java 11 or higher is recommended.

Note: jbang will download and install java from Adopt OpenJDK if no java is available.

Once you have installed from one of the below approaches it is recommended you run jbang app setup to have it setup your PATH to include jbang app scripts it will on operating systems that supports it setup a j! alias you can use instead of jbang.

SDKMan [linux] / [apple]

To install both java and jbang we recommend sdkman on Linux and OSX.

curl -s "https://get.sdkman.io" | bash # (1)
source ~/.bash_profile # (2)

sdk install java # (3)

Once Java is installed and ready, you install jbang with

sdk install jbang

To test your installation run:

jbang --help

This should print out usage information.

To update run:

sdk update jbang

Chocolatey [windows]

On Windows you can install both java and jbang` with Chocolatey.

From a command prompt with enough rights to install with choco:

choco install jdk11

Once Java in installed run:

choco install jbang

To upgrade to latest version:

choco upgrade jbang

The latest package will be published to jbang choco package page, it might be a bit delayed as the review is still manual. In case the default version is not the latest you can see the version list and install specific version using:

choco install jbang --version=<version number>

Scoop [windows]

On Windows you can also install jbang with Scoop.

scoop bucket add jbangdev https://github.com/jbangdev/scoop-bucket
scoop install jbang

To upgrade to latest version:

scoop update jbang

Homebrew [apple]

On OSX you can install 'java' and jbang with Homebrew using custom taps.

To install Java 11:

brew tap AdoptOpenJDK/openjdk
brew cask install adoptopenjdk11

Once Java is installed you can use brew with jbangdev/tap to get jbang:

brew install jbangdev/tap/jbang

To upgrade to latest version:

brew upgrade jbangdev/tap/jbang

(Experimental) Linux packages [linux]

⚠️	These builds are not fully automated yet thus might be slightly behind.

You can install rpm packages from Fedora Copr by doing the following:

dnf copr enable maxandersen/jbang
dnf install jbang

The COPR currently includes builds from various versions of CentOS, Fedora, Mageia and OpenSuse.

Docker / GitHub Action [docker]

You can run jbang via Docker:

docker run -v `pwd`:/ws --workdir=/ws jbangdev/jbang-action helloworld.java

or if you prefer using Quay.io:

docker run -v `pwd`:/ws --workdir=/ws quay.io/jbangdev/jbang-action helloworld.java

The same container images can be used with GitHub Actions, see jbang-action for details.

Manual install

Unzip the latest binary release, add the jbang-<version>/bin folder to your $PATH and you are set.

Wrapper install

If you would like to have jbang available in a local directory and committed into a source code repository (akin to Maven and Gradle wrappers) you can use the jbang wrapper command.

If you have jbang already installed you call jbang wrapper install in a folder to install a local jbang that will run out of that directory using ./jbang.

The ./.jbang directory which jbang wrapper install creates is just a cache which you typically would not commit to a source code repository, so you can e.g. echo .jbang/ >>.gitignore.

"Zero" Install

If you want to try out jbang without a package manager or similar you can run the following to download jbang in ~/.jbang and if necessary java.

Linux/OSX/Windows Bash:

curl -Ls https://sh.jbang.dev | bash -s - <arguments>

For example curl -Ls https://sh.jbang.dev | bash -s - properties@jbangdev

Windows Powershell:

iex "& { $(iwr -useb https://ps.jbang.dev) } <arguments>"

For example iex "& { $(iwr -useb https://ps.jbang.dev) } properties@jbangdev"

Usage

A script is just a single .java file with a classic static main method or a .jsh file which will be passed to jshell.

Below is an (almost) minimal example you can save in helloworld.java or simply run jbang init helloworld.java:

///usr/bin/env jbang "$0" "$@" ; exit $? # (1)

class helloworld { // (2)

    public static void main(String[] args) {
        if(args.length==0) {
            System.out.println("Hello World!");
        } else {
            System.out.println("Hello "   args[0]);
        }
    }
}

By using this // style instead of shebang #! you trick bash, zsh etc. to run this as a script while still being valid java code.
A classname, can be anything when using jbang but to be valid java for most IDEs you’ll want to name it the same as the source file.

Now to run this you can call it via jbang:

jbang helloworld.java

or if on Linux/OSX run it directly. If you created it manually you need to mark it as executable before running it.

chmod  x helloworld.java
./helloworld.java jbang!

⚠️

When no JDK version is available in the PATH, JDK 11 will be downloaded by default to bootstrap jbang. If your script requires a higher version and you don' want to download two JDK’s, you can define an alternative default with JBANG_DEFAULT_JAVA_VERSION env variable.

$ JBANG_DEFAULT_JAVA_VERSION=14 jbang my-script.java

Note that if JDK is found in the PATH, JBANG_DEFAULT_JAVA_VERSION will be ignored.

Default application

If you pass a directory or a url ending in / jbang will look for main.java to run as default application for that directory / location.

URLs from Trusted Sources

You can use http(s):/ and file:/ url’s for input:.

jbang https://gist.github.com/maxandersen/f43b4c52dfcfc42dcd59a04e49acf6ec

For safety reasons jbang will not run arbitrary urls before you indicated you trust their source. Thus when running the above for the first time you will see the following warning about the url not being a trusted source:

jbang https://gist.github.com/maxandersen/f43b4c52dfcfc42dcd59a04e49acf6ec
[jbang] https://gist.github.com/maxandersen/f43b4c52dfcfc42dcd59a04e49acf6ec is not from a trusted source thus not running it automatically.

If you trust the url to be safe to run you can do one of the following:
0) Trust once: Add no trust, just run this time
1) Trust this url in future:
    jbang trust add https://gist.github.com/maxandersen/


Any other response will result in exit.

[jbang] Type in your choice (0 or 1) and hit enter. Times out after 10 seconds.

You can then choose 0 to run once or 1 to trust the suggested url. If you don’t answer within 10 seconds jbang will exit.

To enable running it without such question you need to mark that url or a sub part of it as a trusted source. i.e. jbang trust add https://github.com/maxandersen/ will tell jbang to trust anything with that base url.

You can see more in the comments of the ~/.jbang/trusted-sources.json.

💡	Sites such as GitHub, gitlab, bitbucket, gist, carbon.now.sh jbang will try and extract the proper source rather than the raw html. i.e. doing `jbang https://carbon.now.sh/ae51bf967c98f31a13cba976903030d5 carbon!` is possible.

💡	URL’s will follow redirects. In case you need to use it with sites with self-signed/non-trusted certificates you can if you trust the site use `--insecure`.

Build and run native image (Experimental)

There is support for using native-image from GraalVM project to produce a binary executable.

Since not all java libraries can automatically be built with native-image - especially if using reflection feature are considered highly experimental.

Just run jbang --native helloworld.java and jbang will use native-image from either $JAVA_HOME/bin or $GRAALVM_HOME/bin or $PATH to produce a native image binary.

You can install the native-image utility binary e.g. by installing GraalVM from https://www.graalvm.org/downloads, and then once running gu install native-image as per https://www.graalvm.org/reference-manual/native-image.

💡	If you use `--native` with picocli remember to add `info.picocli:picocli-codegen` as that will ensure it will actually work with `native-image`.

Using `.jsh` for `jshell`

There is support to run .jsh via jshell. The advantage of jshell is that you do not need to have a class or static main method.

Classic jshell does not support passing in arguments nor system properties, jbang does.

In the case of .jsh files jbang injects a startup script that declares a String[] args which will contain any passed in arguments, and it sets any properties passed in as -Dkey=value as parameters to jbang.

That means you can run a script as jbang -Dkey=value helloworld.jsh World and retrieve arguments and properties as:

System.out.println("Hello "   (args.length>0?args[0]:"World")); // (1)
System.out.println(System.getProperty("key")); // (2)

Line where args are accessible without previous declaration.
System properties set when passed as -D arguments to jbang

The script will have the output:

Hello World
value

Please note that .jsh files are source only, they are not compiled thus they are not cached nor can they be built as native images.

💡	If you use `-Dkey` where no value is specified `jbang` will interpret this as `-Dkey=true` allowing you to easily have flags passed into, i.e. `jbang -DskipTests mytestrunner.java`. Now within your script `Boolean.getBoolean('skipTests') will return true.

Running script from standard input

jbang can run scripts directly from standard input using - or /dev/stdin as input.

i.e.

echo 'System.out.println("Hello World!");' | jbang -

💡	If you use `--interactive` `jbang` will let `jshell` enter into interactive/REPL mode. You can write `/exit` to leave this mode.

💡

If your own code needs to handle chained pipes well it is recommended to add the following code:

import sun.misc.Signal;

if (!"Windows".equals(System.getProperty("os.name"))) {
    Signal.handle(new Signal("PIPE"), (final Signal sig) -> System.exit(1));
}

It will give a compiler warning as it is internal API; but for now it works.

Running `.jar`'s

jbang will also run .jar files directly.

i.e. jbang helloworld.jar will run helloworld.jar if found on your local file system.

The .jar can be a local file or a http/https url.

You can also run a .jar file referenced by a Maven coordinate, i.e.:

jbang info.picocli:picocli-codegen:4.5.0

This will fetch the dependency stated and put the transitive dependencies on the class-path.

If you need to specify a main class you can do so by using --main i.e.

jbang --main picocli.codegen.aot.graalvm.ReflectionConfigGenerator info.picocli:picocli-codegen:4.5.0

💡	A side effect of running GAV as a jar, the GAV could also be a `.java` or `.jsh` file and it would be launched as a script instead of a jar. No one would want to do that (right?) but now you know.

Init templates

To get started you can run jbang init helloworld.java and a simple java class with a static main is generated.

Using jbang init --template=cli helloworld.java you get a more complete Hello World CLI using picocli as dependency.

Declare dependencies

If you want to write real scripts you will want to use some java libraries. To specify dependencies you use gradle-style locators or links to Git sources. Below are examples for log4j.

Using `//DEPS`

///usr/bin/env jbang "$0" "$@" ; exit $?
// (1)
//DEPS log4j:log4j:1.2.17

import static java.lang.System.out;

import org.apache.log4j.Logger;
import org.apache.log4j.BasicConfigurator;

import java.util.Arrays;

class classpath_example {

	static final Logger logger = Logger.getLogger(classpath_example.class);

	public static void main(String[] args) {
		BasicConfigurator.configure(); // (2)
		logger.info("Welcome to jbang");

		Arrays.asList(args).forEach(arg -> logger.warn("arg: "   arg));
		logger.info("Hello from Java!");
	}
}

//DEPS must be placed at the start of line and can be one or more space separated dependencies.
Minimal logging setup - required by log4j.

Now when you run this, the first time with no existing dependencies installed you should get an output like this:

$ ./classpath_example.java
[jbang] Resolving dependencies...
[jbang]     Resolving log4j:log4j:1.2.17...Done
[jbang] Dependencies resolved
0 [main] INFO classpath_example  - Welcome to jbang
1 [main] INFO classpath_example  - Hello from Java!

Using links to Git sources

Instead of gradle-style locators you can also use URLs to projects on GitHub, GitLab or BitBucket. Links to those projects will then be converted to artifacts references on jitpack. You can use links to the root of the project, to the root of a tag/release and to specific commits.

If the project you link to has multiple modules and you want only a specific module you can specify the name of the module by appending #name-of-module to the URL.

And finally if the link you provide is to a specific branch of the project then you need to append #:SNAPSHOT to the URL. (If you have both a branch and a module name then use #name-of-module:SNAPSHOT)

Table 1. Examples of links and their resulting locator:

Link	Locator
https://github.com/jbangdev/jbang	com.github.jbangdev:jbang:HEAD-SNAPSHOT
https://github.com/jbangdev/jbang/tree/v1.2.3	com.github.jbangdev:jbang:v1.2.3
https://github.com/jbangdev/jbang/tree/f1f34b031…	com.github.jbangdev:jbang:f1f34b031d
https://github.com/jbangdev/jbang#mymodule	com.github.jbangdev.jbang:mymodule:HEAD-SNAPSHOT
https://github.com/jbangdev/jbang/tree/mybranch#:SNAPSHOT	com.github.jbangdev:jbang:mybranch-SNAPSHOT
https://github.com/jbangdev/jbang/tree/mybranch#mymodule:SNAPSHOT	com.github.jbangdev.jbang.mymodule:mybranch-SNAPSHOT

Offline mode

In case you prefer jbang to just fail-fast when dependencies cannot be found locally you can run jbang in offline mode using jbang -o or jbang --offline. In this mode jbang will simply fail if dependencies have not been cached already.

Repositories

By default jbang uses jcenter as its repository as it is a superset of Maven Central and supposedly should be faster.

And if you are using the above mentioned URL dependencies jitpack will be added automatically as well.

If that is not sufficient for you or need some custom repo you can use //REPOS id=repourl to state which repository URL to use.

For ease of use there are also a few shorthands to use popular commonly available repositories.

Short name	Description
`mavencentral`	Maven Central
`jcenter`	`https://jcenter.bintray.com/`
`google`	`https://maven.google.com/`
`jitpack`	`https://jitpack.io/`

Following example enables use of Maven Central and add a custom acme repository.

//REPOS mavenCentral,acme=https://maven.acme.local/maven

⚠️	If you add any `//REPOS` lines `jbang` will no longer consult `jcenter` thus you need to explicitly add it if needed.

💡	For secure authentication `jbang` will honor `~/.m2/settings-security.xml` for configuring username/passwords.

Using `@Grab`

There is also support for using Groovy lang style @Grab syntax.

///usr/bin/env jbang "$0" "$@" ; exit $?

import static java.lang.System.out;

import org.apache.log4j.Logger;
import org.apache.log4j.BasicConfigurator;

import java.util.Arrays;

import groovy.lang.Grab; // (1)
import groovy.lang.Grapes;
import groovy.lang.GrabResolver;

@GrabResolver("mavenCentral") // (2)
@GrabResolver(name='acme', root='https://maven.acme.local/maven')
@Grapes({ // (3)
		@Grab(group="org.codehaus.groovy", module="groovy", version="2.5.8"), // (4)
		@Grab(module = "log4j", group = "log4j", version = "1.2.17")
})
class classpath_example {

	static final Logger logger = Logger.getLogger(classpath_example.class);

	public static void main(String[] args) {
		BasicConfigurator.configure();
		Arrays.asList(args).forEach(arg -> out.println(arg));
	}
}

Import needed to make the compiler be okey with @Grab annotation.
Using GrabResolver to enable mavenCentral and custom acme repository
In Groovy you normally put @Grab on import statements. That is not allowed in Java thus when having multiple imports you need to put them in a @Grapes annotation first.
jbang will grab any @Grab annotation and assume it is declaring dependencies.

System properties and Environment variables

In dependencies you can refer to environment and system properties to parameterize the dependencies. It uses the format ${[env.]propertyname:<defaultvalue>}.

Furthermore to align with properties commonly used to make dependency resolution portable jbang exposes properties similar to what the os-maven-plugin does. Plus for ease of use for javafx dependencies it also setups a property named ${os.detected.jfxname}.

Examples:

${env.USER} = 'max'
${os.name} = 'Mac OS X'
${non.existing.key:empty} = 'empty'
${os.detected.jfxname} = 'mac'

This can be put to use in //DEPS like so:

//DEPS org.openjfx:javafx-graphics:11.0.2:${os.detected.jfxname}

Here we use the properties to avoid hardcoding your script to a specific operating system.

JavaFX

If jbang detects you have a javafx- dependency in your list of dependencies it will if you java command supports Java modules automatically set the necessary --module-path and --add-modules.

See examples/jfx.java and examples/jfxtiles.java for examples of this.

Multiple source files

It is possible to use multiple source files just by having the files in the same source directory, it will even work for a limited extent with packages.

For example the example below works by just calling jbang Main.java:

Main.java

import model.Person;

public class Main {

    public static void main(String... args) {
        Person p = new Person(args[0]);
        System.out.println("Hello "   p.getName());
    }
}

model/Person.java

package model;

public class Person {
    String name;
    public String getName() { return name; }
    public Person(String n) { this.name = n; }
}

There are some cases where the above does not work; i.e. if two packages refer to each other - i.e. model.Person referring to util.Generator will fail. Also jbang edit does not know about multiple sources as it runs and must run before compilation occurs.

Thus version 0.46 there is now support for having that all work with multiple source files. The main script file defines all the dependencies and you add more source files into the application using //SOURCES <filename>. If included source has //SOURCES that will also get included recursively.

The listed file name(s) gets added to source list when compiling.

Currently there are not *.java style matching or support for these .java files to declare //DEPS or other jbang configuration. That will currently only be honored by the main script/app. These will be loosened up in future based on feedback.

Adding more resources

If you want to add a META-INF/application.properties or META-INF/resource.index.html or other files to the generated jar you can use //FILES to add them.

The format is //FILES <mountpoint>[=<sourcefile>].

Example:

//FILES resource.properties
//FILES META-INF/resources/index.html=index.html

Here resource.properties will be copied as is and META-INF/resources/index.html gets its content from index.html.

All locations are relative to the script location.

⚠️	Currently `jbang edit` and http(s) based script do not work with `//FILES`. Will be added later.

Extension-less/non-java files for cli-plugins

You can use jbang to write plugins for cli’s like kubectl, git, etc. They expect their plugins to be named like <cmd>-<plugin>, i.e. kubectl-myplugin.

Furthermore some of them, particularly kubectl currently require the file to start with #! otherwise you get a exec format error.

There are two ways to have that work. The first recommended way is to use jbang app install which setups an intermediate script to avoid the issue, i.e. jbang app install --name kubectl-my-plugin myplugin.java.

The second is to use a bit of auto-magic jbang has to help in case you only want a single file, no intermediate script. That is described below.

Kebab casing of filename

jbang lets you name your file without a .java or .jsh extension, such as kubectl-my-plugin or myjavascript.sh. jbang will in this case copy the file to a temporary directory using kebab-case to map the name to a proper java class name.

For example, if you make a file called kubectl-my-plugin then jbang will assume the actual class name to launch to be KubectlMyPlugin.

Note, similar is done when using jbang edit, here the symbolic link will be made so the IDE will treat it as regular camel cased java class.

📎	If you do not follow this naming pattern you will get a compile error as `javac` expects both the public class and file names to be equal.

she-bang auto-stripped

For extension less scripts, you can put #!' header at the beginning to let apps recognize it is to be treated as a script. To avoid issues when compiling, `jbang will remove that line before compilation.

For now this is required for kubectl plugin but not git. Issue opened on this limitation.

Java version

jbang will by default use JAVA_HOME and if not available, check the PATH to locate the java executable to run the script with.

If your script requires a specific or minimal version of Java you can use //JAVA <version>( ).

If Jbang finds a java executable using JAVA_HOME or PATH which satisfies the stated java version jbang will use it. If no such version is found it will automatically download and install it.

Examples:

//JAVA 11 will force use of Java 11.
//JAVA 13 will require at least java 13. Java 13 or higher will be used.

In case no matching java is found jbang will fail.

You can always force running with specific version of java using --java command line option, i.e. jbang --java 8 hello.java

Managing JDKs

In the previous section it was mentioned that Jbang will automatically download and install JDKs when necessary. You can use the jdk command to manage JDKs, for example you can run the following:

jbang jdk list

which will list all the JDKs that are currently installed by Jbang.

It’s easy to install additional JDKs by running:

jbang jdk install 14

which will download and install JDK version 14 into Jbang’s cache (~/.jbang/cache/jdks by default). The list of versions that are available for installation can be found here: https://adoptopenjdk.net/releases.html

The first JDK that gets installed by Jbang will be set as the "default" JDK. This is from then on the JDK that will be used by Jbang if no Java could be found on the system (meaning javac wasn’t found on the PATH and no JAVA_HOME is set). You can change the default JDK by running:

jbang jdk default 12

Running it without an argument will return the version of the JDK that is currently set as the default.

📎	On Windows you might need elevated privileges to create symbolic links. If you don’t have permissions then running the above command will result in an error. To use it enable symbolic links for your user or run your shell/terminal as administrator to have this feature working.

When you uninstall a JDK by running:

jbang jdk uninstall 12

and that JDK was set as the default, Jbang will set the next higher version JDK as the default. If no higher version is available it will select the next lower version.

Using managed JDKs yourself

Given the fact that Jbang is able to easily download and install JDKs we thought that it might be a good option for our users to be able to access those JDKs for their own use instead of having to install yet another version themselves.

To make that easy we added a couple of useful commands. The first can be used to set retrieve to location where the JDK is installed:

jbang jdk home

This will return the path to the "default" JDK (by default ~/.jbang/currentjdk), if you want to know the location of a specific JDK you can pass the version as an argument: jbang jdk home 14. This command could be used by scripts to find a JDK to use to run a Java program for example (eg: JAVA_HOME=$(jbang jdk home).

For setting up your current command line environment there’s something simpler. You can run:

jbang jdk java-env

On Linux and Mac this will output something like:

export PATH="/home/user/.jbang/currentjdk/bin:$PATH"
export JAVA_HOME="/home/user/.jbang/currentjdk"
# Run this command to configure your shell:
# eval $(jbang jdk java-env)

And the output itself shows how to properly use it to configure your command line to use the JDK. In this case it’s by running:

eval $(jbang jdk java-env)

To do this by default for all shells you start simply add the above line to your ~/.bashrc file.

Unfortunately on Windows using CMD things are not as easy as is show by the output of jbang jdk java-env on that platform:

set PATH=C:\Users\user\.jbang\currentjdk\bin;%PATH%
set JAVA_HOME=C:\Users\user\.jbang\currentjdk
rem Copy & paste the above commands in your CMD window or add
rem them to your Environment Variables in the System Settings.

Instead of copying and pasting lines you could also redirect the output to a .bat file and execute that instead:

> jbang jdk java-env > setenv.bat
> setenv

Editing

You can edit your script in an IDE/editor by using jbang edit helloworld.java. This will generate a project in a temporary location with symbolic links to your script and output the generated folder name. The easiest way to use that is to use it in a call to your IDE:

code `jbang edit helloworld.java`

If you add further dependencies to your file just re-run the edit command and the relevant files will be regenerated with the updated dependencies.

Above does require using a shell that allows for variable evaluation, if you are on i.e. Windows then you might prefer using:

jbang edit --open=[editor] helloworld.java

The editor used will be what is specified as the argument to --open or value of JBANG_EDITOR environment variable. The editor command must be available on the PATH to be executed from jbang. If you are executing jbang edit --open=code helloworld.java a code executable (visual studio code) must be on the PATH. Next to this you can pass the full path to the open parameter like in --open=/usr/bin/code.

📎	On Windows you might need elevated privileges to create symbolic links. If you don’t have permissions then the `edit` option will result in an error. To use it enable symbolic links for your user or run your shell/terminal as administrator to have this feature working.

Install of default vscodium editor [EXPERIMENTAL]

If no editor available at all jbang will install VSCodium (free/libre version of Visual Studio code) with default java extensions enabled in so called "portable mode".

This automatic install and setup of editor is fully optional and if you have another IDE or editor already installed use it using jbang edit --open=<editor> or set JBANG_EDITOR environment variable to have jbang use it by default.

Live Editing

You can also use jbang edit --live and jbang will launch your editor while watching for file changes and regenerate the temporary project to pick up changes in dependencies.

IDE and Editor support

The edit feature been tested with the following IDE’s:

Table 2. IDE’s and Editors tested with jbang

Neovim w/ spacevim Java

Emacs w/ Spacemacs Java

The edit feature works with various IDE’s - it generates a build.gradle to use with IDE’s that understands Gradle directly. For speed and consistency jbang also generates IDE specific settings.

Currently launchers and project files are generated for Eclipse and vscode. Intellij just reads build.gradle for now thus to run/debug you will need to manually set it up.

Some editors like Code or IDEA are not per default executable / startable from the command line. Here are some helpfull documentations for supporting command line usage of different editors:

Debugging

When running .java scripts with jbang you can pass the --debug-flag and the script will enable debug, suspend the execution and wait until you connect a debugger to port 4004.

jbang --debug helloworld.java
Listening for transport dt_socket at address: 4004

You can change the debug port and host by passing in a interface pattern and number to the debug argument, e.g., --debug=*:4321.

This will make it use port 4321 and make it listen on all ('*') network interfaces.

📎	Be sure to put a breakpoint in your IDE/debugger before you connect to make the debugger actually stop when you need it.

Debug jbang itself

Java itself will add JAVA_TOOL_OPTIONS which will apply to jbang too.

For finer and more explicit control the scripts, jbang will add JBANG_JAVA_OPTIONS to the call to jbang itself. Thus if you want to enable debug or other details for jbang set that environment variable.

Flight Recorder

Flight recorder is a feature of the Java VM that lets you gather diagnostic and profiling data about your script.

You can use //JAVA_OPTIONS to have full control over it; but for the easiest setup jbang lets you just run with --jfr, e.g.,

jbang --jfr myapp.java

By default --jfr will start flight recorder and tell it to dump event recordings to myapp.jfr (i.e. using base name of the script as its filename).

Then you can use tools like jvisualvm or jmc to explore the data.

If you want to tweak the configuration you can pass flight recorder options, like jbang --jfr=filename={baseName}.jfr,maxage=24h where {baseName} will be replaced by the filename and then added maxage=24h to flight recording options.

If you want further control use //JAVAC_OPTS -XX:StartFlightRecording=<your options> instead.

`java` and `javac` Options

If you want to tweak memory settings or enable preview features you can setup the necessary options using //JAVA_OPTS and //COMPILER_OPTS as in the following example using Java 14 experimental record feature:

///usr/bin/env jbang "$0" "$@" ; exit $?
//JAVAC_OPTIONS --enable-preview -source 14 (1)
//JAVA_OPTIONS --enable-preview // (2)

import static java.lang.System.*;

public class records {

    record Point(int x, int y) {}

    public static void main(String[] args) {
        var p = new Point(2,4);
        out.println(p);
    }
}

Since Java 9 JDK_JAVA_OPTIONS and JDK_JAVAC_OPTIONS are also picked up by the Java runtime and compiler automatically.

For Java 8 and if you want to set explicitly only for jbang you can also add flags by setting JBANG_JAVA_OPTIONS and JBANG_JAVAC_OPTIONS respectively.

If your scripts uses a lot of classes Class Data Sharing might help on your startup. The following requires Java 13 .

Using --cds jbang will build the jar with Application Class Data Sharing enabled and when run have it load shared class data.

You can put //CDS in the java file to enable it by default, or simply use --cds to force it or --no-cds to turn it off no matter what the jbang script file contains.

Aliases

To avoid remembering long paths and to enable easy launch of jbang scripts there is an alias command to setup and manage aliases to actual scripts.

jbang alias add hello https://github.com/jbangdev/jbang-examples/blob/master/examples/helloworld.java

will add an alias named hello pointing to that github url which then can be run using jbang hello.

jbang alias list

will show you all the aliases that are defined locally.

Implicit Alias Catalogs

The aliases you create are stored locally (see Local Alias Catalogs), but Jbang can also use remote catalogs. You can access those catalogs explicitly (see Catalogs) but it is much easier to use what we call "implicit catalogs", which are aliases that have a special format and Jbang is smart enough to know where to find their definition.

Examples:

jbang hello@jbangdev will run the alias hello as defined in jbang-catalog.json found in https://github.com/jbangdev/jbang-catalog.

This allows anyone to provide a set of jbang scripts defined in their github, gitlab or bitbucket repositories.

The full format is <alias>@<user/org>(/repository)(/branch)(~path) allowing you to do things like:

Table 3. Implicit Catalog Examples:

Command	Description
`jbang hello@acme`	`hello` alias found in `acme/jbang-catalog/jbang-catalog.json` of the default branch searched on github, gitlab and bitbucket in that order.
`jbang hello@acme/mycatalog`	`hello` found in `acme/mycatalog/jbang-catalog.json` searched on github, gitlab and bitbucket in that order.
`jbang hello@acme/mycatalog/dev`	`hello` found in `acme/mycatalog/jbang-catalog.json` in the `dev` branch searched on github, gitlab and bitbucket in that order.
`jbang hello@acme~experimental`	`hello` found in `acme/jbang-catalog/experimental/jbang-catalog.json`

Local Alias Catalogs

Jbang will also look in the current directory for a jbang-catalog.json file and if it exists it will look up any aliases in there too. In fact it will look in several places in the following order:

Current directory, ./jbang-catalog.json
In .jbang/jbang-catalog.json
In the parent directory, ../jbang-catalog.json
In the parent’s .jbang directory, ../.jbang/jbang-catalog.json`
And repeating steps 3 and 4 recursively upwards to the root of the file system
As the last step it will look in $HOME/.jbang/jbang-catalog.json

Jbang will use any aliases defined in those files, but on top of that it will also look at the aliases defined in any catalogs mentioned in those files as well. Aliases defined in the file have preference over aliases found in any catalogs defined in the same file.

When you create aliases using jbang alias add, or add catalogs using jbang catalog add the same ordering will be used to determine where to store the alias or catalog. Btw, this will only take into account existing files! So if no jbang-catalog.json file exists in the local directory it will not be created for you, but Jbang will keep looking until it finds a file to use (as a last option it will always be written to $HOME/.jbang/jbang-catalog.json).

This means that if you want to write the alias to jbang-catalog.json in your local folder you will either have to create the file first (eg by running touch jbang-catalog.json) or by explicitly specifying the file location:

jbang alias add -f jbang-catalog.json hello https://github.com/jbangdev/jbang-examples/blob/master/examples/helloworld.java

Btw, the flag --show-origin is very useful when listing aliases to find out where exactly an alias is defined:

jbang alias list --show-origin

Catalogs

Catalogs are lists of Aliases as defined in the previous section, but while the alias command is used to manage aliases within a catalog, the catalog command is for managing references to catalogs. This is mostly useful when dealing with remote catalogs. You can for example add a catalog like this:

jbang catalog add demo https://github.com/jbangdev/jbang-catalog/blob/master/jbang-catalog.json

or simply by using the same "implicit" catalog system described in Implicit Alias Catalogs:

jbang catalog add demo jbangdev

The aliases in that catalog are now available by adding @demo to their names. For example:

$ jbang alias list demo
env@demo = Dump table of Environment Variables
gavsearch@demo = Search search.maven.org for maven artifacts.
hello@demo = Script that says hello back for each argument
properties@demo = Dump table of System properties
$ jbang run hello@demo World!
[jbang] Building jar...
Hello World!

In fact it’s possible to run the alias just by using jbang run hello, the @demo part is only necessary when trying to disambiguate between aliases with the same name from different catalogs.

You can list the available catalogs by running:

jbang catalog list

NB: The output will not only show the catalogs you defined yourself but also the ones that get added implicitly when running aliases as described in the section Implicit Alias Catalogs.

Java Agents

You can activate a javaagent using --javaagent=<agent>[=<options>] where agent can be a already packaged agent jar from file, http url or Maven Coordinate.

It can also be a jbang script itself where you have put //JAVAAGENT to activate agent packaging.

You can create a basic agent using jbang init -t agent myagent.java to get started.

Bash/Zsh auto-completion

If you are using bash or zsh in your terminal you can get auto-completion by running the following:

source <(jbang completion)

Caching

In previous versions of jbang, Java 10 direct launch of .java was used, but since v0.6 jbang works with Java 8 and thus it needs to do a separate compile step. Besides now working with Java 8 this also allows to cache the compiled script and thus launch faster on consecutive runs.

The caching goes to ~/.jbang/cache by default, you can run jbang cache clear to remove all cache data from this folder.

Install apps [EXPERIMENTAL]

Since version 0.56 jbang comes with ability to setup jbang to put scripts/apps into your PATH using jbang app. This is useful to easily make scripts available from anywhere on any Operating System;

To get started run jbang app setup this will on Windows modify your system wide PATH to include a folder managed by jbang. On bash/zsh based shells it will setup PATH in your bashrc or zshrc file(s.)

Once setup, you can use jbang app install <scriptRef> to install that script into the jbang managed path.

i.e. jbang app install myscript.java will add myscript as a command you can run.

You can also use aliases, like jbang app install gavsearch@jbangdev.

If you have two script/apps with same name or just want to use a specific name you can use --name to control the generated command: jbang app install --name mvnsearch gavsearch@jbangdev

If you want to see which are already installed use jbang list and you can use jbang uinstall <name> to uninstall the script/app.

Build Integration [EXPERIMENTAL]

While jbang prepares and builds the underlying jar used for launch there is since v0.40 (for now) experimental API allowing user included dependencies to influence the generated jar and possible native image.

An example use case enabled by this is to have full Quarkus integration, jbang quarkuscode.java will have Quarkus participate to perform its build time optimizations rather than doing it at runtime every time.

It works as following:

Before the jar is created jbang will scan the classpath for META-INF/jbang-integration.list. Any classes listed in this file will be loaded and jbang will expect and call the following method on these classes:

/**
*
* @param param build dir directory which will be made into a jar when build is done
* @param pomFile location of pom.xml representing the projects dependencies
* @param dependencies list of GAV to Path of artifact/classpath dependencies
* @param nativeImage true if --native been requested
* @return Map<String, Object> map of returns; special keys are "native-image" which is a and "files" to
*          return native-image to be run and list of files to get written to the output directory.
*
*/
Map<String, Object> postBuild(Path builddir, Path pomFile, List<Map.Entry<String, Path>> dependencies,
            boolean nativeImage)

Still very experimental and bound to change. Example of its use can be found in Quarkus.

Cheatsheet

You can get examples on running/using jbang using cheat command with cheat jbang.

FAQ

Why the name j’bang?

I was reading up on how to use the new shebang (#!) feature support in Java 10 and came up with the idea of port kscript to Java and needed a name. From there came j’bang which is a "bad" spelling of how shebang is pronounced in French.
Why use gradle resource locators rather than ?

kscript used it and it’s nice as it is a one-liner and easily parsable.
How does this compare to ?

After doing jbang I’ve learned about similar projects and thought it would be nice with some comparison;

jgo: an alternative way to launch jars using maven coordinates. Implemented in python, depends on Java and Maven to be available. Not really for scripting but a novel way to launch java apps already packaged as a maven dependency.
Why would I use Java to write scripts ? Java sucks for that… Use groovy, kotlin, scala, etc. instead!

Well, does it really suck ? With Java 8 streams, static imports and greatly improved standard java libraries it is very close to what kscript and grape look like. With the following advantages:
- works with plain Java without installing additional compiler/build tools
- all IDE’s support editing .java files very well, content assist, etc.
- great debugging
  
  And to be honest I built jbang just to see if I could and get my Java skills refreshed for the newer features in the language. Use it at your own risk :)
Why not use normal shebang(#!) in the header ?

You can use normal shebang (#!/usr/bin/env jbang) and Java 10 will actually work with it from the command line. Not recommended though as many tools and especially IDE’s will start complaining about syntax errors as they don’t ignore the first line in this case.

By using the // form it is treated as both a bash/shell file AND a valid java file and thus works everywhere a java file will work.

It’s worth noting that Go uses a similar approach which is also where I learned it from.
HELP! My code formatter keeps breaking my // directives!

When using automated code formatting tools, some care and configuration must be made to prevent the tooling from rewriting and preventing jbang from working as expected.

Use the following configuration blocks to correctly configure your tool:
Table 4. Configuration Tool Settings:
Formatting Tool Configuration
Clang Format

CommentPragmas: '^[^ ]'

Thanks

jbang was heavily inspired by how kscript by Holger Brand works.

Name		Name	Last commit message	Last commit date
Latest commit History 933 Commits
.github		.github
examples		examples
gradle/wrapper		gradle/wrapper
images		images
itests		itests
misc		misc
src		src
.all-contributorsrc		.all-contributorsrc
.gitattributes		.gitattributes
.gitignore		.gitignore
.gitpod.Dockerfile		.gitpod.Dockerfile
.gitpod.yml		.gitpod.yml
CONTRIBUTING.adoc		CONTRIBUTING.adoc
CONTRIBUTORS.md		CONTRIBUTORS.md
LICENSE		LICENSE
build.gradle		build.gradle
devfile.yaml		devfile.yaml
gradlew		gradlew
gradlew.bat		gradlew.bat
jbang-catalog.json		jbang-catalog.json
otp.java		otp.java
pipelines.puml		pipelines.puml
readme.adoc		readme.adoc
settings.gradle		settings.gradle

License

msohn/jbang

Folders and files

Latest commit

History

Repository files navigation

jbang - Unleash the power of Java

Intro

Features

Requirements

Installation

SDKMan [linux] / [apple]

Chocolatey [windows]

Scoop [windows]

Homebrew [apple]

(Experimental) Linux packages [linux]

Docker / GitHub Action [docker]

Manual install

Wrapper install

"Zero" Install

Usage

Default application

URLs from Trusted Sources

Build and run native image (Experimental)

Using .jsh for jshell

Running script from standard input

Running .jar's

Init templates

Declare dependencies

Using //DEPS

Using links to Git sources

Offline mode

Repositories

Using @Grab

System properties and Environment variables

JavaFX

Multiple source files

Adding more resources

Extension-less/non-java files for cli-plugins

Kebab casing of filename

she-bang auto-stripped

Java version

Managing JDKs

Using managed JDKs yourself

Editing

Install of default vscodium editor [EXPERIMENTAL]

Live Editing

IDE and Editor support

Debugging

Debug jbang itself

Flight Recorder

java and javac Options

(Experimental) Application Class Data Sharing

Aliases

Implicit Alias Catalogs

Local Alias Catalogs

Catalogs

Java Agents

Bash/Zsh auto-completion

Caching

Install apps [EXPERIMENTAL]

Build Integration [EXPERIMENTAL]

Cheatsheet

FAQ

Thanks

About

Resources

License

Stars

Watchers

Forks

Releases

Packages 0

Languages

Using `.jsh` for `jshell`

Running `.jar`'s

Using `//DEPS`

Using `@Grab`

`java` and `javac` Options

Packages