Créer un programme de ligne de commande Java avec Picocli

1. Introduction

Dans ce tutoriel, nous aborderons la bibliothèque picocli , qui nous permet de créer facilement des programmes en ligne de commande en Java.

Nous allons commencer par créer une commande Hello World. Nous allons ensuite approfondir les fonctionnalités clés de la bibliothèque en reproduisant, partiellement, la commande git .

2. Hello World Command

Commençons par quelque chose de simple: une commande Hello World!

Tout d'abord, nous devons ajouter la dépendance au projet picocli :

 info.picocli picocli 3.9.6 

Comme nous pouvons le voir, nous utiliserons la version 3.9.6 de la bibliothèque, bien qu'une version 4.0.0 soit en construction (actuellement disponible en test alpha).

Maintenant que la dépendance est configurée, créons notre commande Hello World. Pour ce faire, nous utiliserons l' annotation @Command de la bibliothèque :

@Command( name = "hello", description = "Says hello" ) public class HelloWorldCommand { }

Comme on peut le voir, l'annotation peut prendre des paramètres. Nous n'en utilisons que deux ici. Leur objectif est de fournir des informations sur la commande actuelle et le texte du message d'aide automatique.

Pour le moment, nous ne pouvons pas faire grand-chose avec cette commande. Pour en faire quelque chose, nous devons ajouter une principale méthode d' appel de la commodité CommandLine.run (Runnable, String []) méthode . Cela prend deux paramètres: une instance de notre commande, qui doit donc implémenter l' interface Runnable , et un tableau String représentant les arguments de la commande (options, paramètres et sous-commandes):

public class HelloWorldCommand implements Runnable { public static void main(String[] args) { CommandLine.run(new HelloWorldCommand(), args); } @Override public void run() { System.out.println("Hello World!"); } }

Maintenant, lorsque nous exécutons la méthode main , nous verrons que la console affiche "Hello World!"

Lorsqu'il est emballé dans un fichier jar, nous pouvons exécuter notre commande Hello World à l'aide de la commande java :

java -cp "pathToPicocliJar;pathToCommandJar" com.baeldung.picoli.helloworld.HelloWorldCommand

Sans surprise, cela génère également le "Hello World!" chaîne à la console.

3. Un cas d'utilisation concret

Maintenant que nous avons vu les bases, nous allons plonger dans la bibliothèque picocli . Pour ce faire, nous allons reproduire, partiellement, une commande populaire: git .

Bien sûr, le but ne sera pas d'implémenter le comportement de la commande git mais de reproduire les possibilités de la commande git - quelles sous-commandes existent et quelles options sont disponibles pour une sous-commande particulière.

Tout d'abord, nous devons créer une classe GitCommand comme nous l'avons fait pour notre commande Hello World:

@Command public class GitCommand implements Runnable { public static void main(String[] args) { CommandLine.run(new GitCommand(), args); } @Override public void run() { System.out.println("The popular git command"); } }

4. Ajout de sous-commandes

La commande git offre de nombreuses sous-commandes - add, commit, remote et bien d'autres. Nous allons nous concentrer ici sur l' ajout et le commit .

Donc, notre objectif ici sera de déclarer ces deux sous-commandes à la commande principale. Picocli propose trois façons d'y parvenir.

4.1. Utilisation de l' annotation @Command sur les classes

L' annotation @Command offre la possibilité d'enregistrer des sous-commandes via le paramètre subcommands :

@Command( subcommands = { GitAddCommand.class, GitCommitCommand.class } )

Dans notre cas, nous ajoutons deux nouvelles classes: GitAddCommand et GitCommitCommand . Les deux sont annotés avec @Command et implémentent Runnable . Il est important de leur donner un nom, car les noms seront utilisés par picocli pour reconnaître quelle (s) sous-commande (s) exécuter:

@Command( name = "add" ) public class GitAddCommand implements Runnable { @Override public void run() { System.out.println("Adding some files to the staging area"); } }
@Command( name = "commit" ) public class GitCommitCommand implements Runnable { @Override public void run() { System.out.println("Committing files in the staging area, how wonderful?"); } }

Ainsi, si nous exécutons notre commande principale avec add comme argument, la console affichera «Ajout de fichiers à la zone de préparation» .

4.2. Utilisation de l' annotation @Command sur les méthodes

Une autre façon de déclarer des sous-commandes consiste à créer des méthodes annotées @Command représentant ces commandes dans la classe GitCommand :

@Command(name = "add") public void addCommand() { System.out.println("Adding some files to the staging area"); } @Command(name = "commit") public void commitCommand() { System.out.println("Committing files in the staging area, how wonderful?"); }

De cette façon, nous pouvons directement implémenter notre logique métier dans les méthodes et ne pas créer de classes séparées pour la gérer.

4.3. Ajout de sous-commandes par programme

Enfin, picocli nous offre la possibilité d'enregistrer nos sous-commandes par programmation. Celui-ci est un peu plus délicat, car nous devons créer un objet CommandLine encapsulant notre commande, puis y ajouter les sous-commandes:

CommandLine commandLine = new CommandLine(new GitCommand()); commandLine.addSubcommand("add", new GitAddCommand()); commandLine.addSubcommand("commit", new GitCommitCommand());

Après cela, nous devons encore exécuter notre commande, mais nous ne pouvons plus utiliser la méthode CommandLine.run () . Maintenant, nous devons appeler la méthode parseWithHandler () sur notre objet C ommandLine nouvellement créé :

commandLine.parseWithHandler(new RunLast(), args);

Nous devons noter l'utilisation de la classe RunLast , qui indique à picocli d'exécuter la sous-commande la plus spécifique. Il existe deux autres gestionnaires de commandes fournis par picocli : RunFirst et RunAll . Le premier exécute la commande la plus haute, tandis que le second les exécute tous.

Lorsque vous utilisez la méthode pratique CommandLine.run () , le gestionnaire RunLast est utilisé par défaut.

5. Gestion des options à l'aide de l' annotation @Option

5.1. Option sans argument

Let's now see how to add some options to our commands. Indeed, we would like to tell our add command that it should add all modified files. To achieve that, we'll add a field annotated with the @Option annotation to our GitAddCommand class:

@Option(names = {"-A", "--all"}) private boolean allFiles; @Override public void run() { if (allFiles) { System.out.println("Adding all files to the staging area"); } else { System.out.println("Adding some files to the staging area"); } }

As we can see, the annotation takes a names parameter, which gives the different names of the option. Therefore, calling the add command with either -A or –all will set the allFiles field to true. So, if we run the command with the option, the console will show “Adding all files to the staging area”.

5.2. Option with an Argument

As we just saw, for options without arguments, their presence or absence is always evaluated to a boolean value.

However, it's possible to register options that take arguments. We can do this simply by declaring our field to be of a different type. Let's add a message option to our commit command:

@Option(names = {"-m", "--message"}) private String message; @Override public void run() { System.out.println("Committing files in the staging area, how wonderful?"); if (message != null) { System.out.println("The commit message is " + message); } }

Unsurprisingly, when given the message option, the command will show the commit message on the console. Later in the article, we'll cover which types are handled by the library and how to handle other types.

5.3. Option with Multiple Arguments

But now, what if we want our command to take multiple messages, as is done with the real git commit command? No worries, let's make our field be an array or a Collection, and we're pretty much done:

@Option(names = {"-m", "--message"}) private String[] messages; @Override public void run() { System.out.println("Committing files in the staging area, how wonderful?"); if (messages != null) { System.out.println("The commit message is"); for (String message : messages) { System.out.println(message); } } }

Now, we can use the message option multiple times:

commit -m "My commit is great" -m "My commit is beautiful"

However, we might also want to give the option only once and separate the different parameters by a regex delimiter. Hence, we can use the split parameter of the @Option annotation:

@Option(names = {"-m", "--message"}, split = ",") private String[] messages;

Now, we can pass -m “My commit is great”,”My commit is beautiful” to achieve the same result as above.

5.4. Required Option

Sometimes, we might have an option that is required. The required argument, which defaults to false, allows us to do that:

@Option(names = {"-m", "--message"}, required = true) private String[] messages;

Now it's impossible to call the commit command without specifying the message option. If we try to do that, picocli will print an error:

Missing required option '--message=' Usage: git commit -m= [-m=]... -m, --message=

6. Managing Positional Parameters

6.1. Capture Positional Parameters

Now, let's focus on our add command because it's not very powerful yet. We can only decide to add all files, but what if we wanted to add specific files?

We could use another option to do that, but a better choice here would be to use positional parameters. Indeed, positional parameters are meant to capture command arguments that occupy specific positions and are neither subcommands nor options.

In our example, this would enable us to do something like:

add file1 file2

In order to capture positional parameters, we'll make use of the @Parameters annotation:

@Parameters private List files; @Override public void run() { if (allFiles) { System.out.println("Adding all files to the staging area"); } if (files != null) { files.forEach(path -> System.out.println("Adding " + path + " to the staging area")); } }

Now, our command from earlier would print:

Adding file1 to the staging area Adding file2 to the staging area

6.2. Capture a Subset of Positional Parameters

It's possible to be more fine-grained about which positional parameters to capture, thanks to the index parameter of the annotation. The index is zero-based. Thus, if we define:

@Parameters(index="2..*")

This would capture arguments that don't match options or subcommands, from the third one to the end.

The index can be either a range or a single number, representing a single position.

7. A Word About Type Conversion

As we've seen earlier in this tutorial, picocli handles some type conversion by itself. For example, it maps multiple values to arrays or Collections, but it can also map arguments to specific types like when we use the Path class for the add command.

As a matter of fact, picocli comes with a bunch of pre-handled types. This means we can use those types directly without having to think about converting them ourselves.

However, we might need to map our command arguments to types other than those that are already handled. Fortunately for us, this is possible thanks to the ITypeConverter interface and the CommandLine#registerConverter method, which associates a type to a converter.

Let's imagine we want to add the config subcommand to our git command, but we don't want users to change a configuration element that doesn't exist. So, we decide to map those elements to an enum:

public enum ConfigElement { USERNAME("user.name"), EMAIL("user.email"); private final String value; ConfigElement(String value) { this.value = value; } public String value() { return value; } public static ConfigElement from(String value) { return Arrays.stream(values()) .filter(element -> element.value.equals(value)) .findFirst() .orElseThrow(() -> new IllegalArgumentException("The argument " + value + " doesn't match any ConfigElement")); } }

Plus, in our newly created GitConfigCommand class, let's add two positional parameters:

@Parameters(index = "0") private ConfigElement element; @Parameters(index = "1") private String value; @Override public void run() { System.out.println("Setting " + element.value() + " to " + value); }

This way, we make sure that users won't be able to change non-existent configuration elements.

Finally, we have to register our converter. What's beautiful is that, if using Java 8 or higher, we don't even have to create a class implementing the ITypeConverter interface. We can just pass a lambda or method reference to the registerConverter() method:

CommandLine commandLine = new CommandLine(new GitCommand()); commandLine.registerConverter(ConfigElement.class, ConfigElement::from); commandLine.parseWithHandler(new RunLast(), args);

This happens in the GitCommand main() method. Note that we had to let go of the convenience CommandLine.run() method.

When used with an unhandled configuration element, the command would show the help message plus a piece of information telling us that it wasn't possible to convert the parameter to a ConfigElement:

Invalid value for positional parameter at index 0 (): cannot convert 'user.phone' to ConfigElement (java.lang.IllegalArgumentException: The argument user.phone doesn't match any ConfigElement) Usage: git config    

8. Integrating with Spring Boot

Finally, let's see how to Springify all that!

Indeed, we might be working within a Spring Boot environment and want to benefit from it in our command-line program. In order to do that, we must create a SpringBootApplicationimplementing the CommandLineRunner interface:

@SpringBootApplication public class Application implements CommandLineRunner { public static void main(String[] args) { SpringApplication.run(Application.class, args); } @Override public void run(String... args) { } }

Plus, let's annotate all our commands and subcommands with the Spring @Component annotation and autowire all that in our Application:

private GitCommand gitCommand; private GitAddCommand addCommand; private GitCommitCommand commitCommand; private GitConfigCommand configCommand; public Application(GitCommand gitCommand, GitAddCommand addCommand, GitCommitCommand commitCommand, GitConfigCommand configCommand) { this.gitCommand = gitCommand; this.addCommand = addCommand; this.commitCommand = commitCommand; this.configCommand = configCommand; }

Note that we had to autowire every subcommand. Unfortunately, this is because, for now, picocli is not yet able to retrieve subcommands from the Spring context when declared declaratively (with annotations). Thus, we'll have to do that wiring ourselves, in a programmatic way:

@Override public void run(String... args) { CommandLine commandLine = new CommandLine(gitCommand); commandLine.addSubcommand("add", addCommand); commandLine.addSubcommand("commit", commitCommand); commandLine.addSubcommand("config", configCommand); commandLine.parseWithHandler(new CommandLine.RunLast(), args); }

And now, our command line program works like a charm with Spring components. Therefore, we could create some service classes and use them in our commands, and let Spring take care of the dependency injection.

9. Conclusion

In this article, we've seen some key features of the picocli library. We've learned how to create a new command and add some subcommands to it. We've seen many ways to deal with options and positional parameters. Plus, we've learned how to implement our own type converters to make our commands strongly typed. Finally, we've seen how to bring Spring Boot into our commands.

Of course, there are many things more to discover about it. The library provides complete documentation.

As for the full code of this article, it can be found on our GitHub.