ModularBot - JS Nashorn Support

Module that can load modules in JavaScript.

License

License

Categories

Categories

JavaScript Languages
GroupId

GroupId

com.jesus-crie
ArtifactId

ArtifactId

modularbot-nashorn-support
Last Version

Last Version

2.5.0_23
Release Date

Release Date

Type

Type

jar
Description

Description

ModularBot - JS Nashorn Support
Module that can load modules in JavaScript.
Project URL

Project URL

https://github.com/JesusCrie/ModularBot
Source Code Management

Source Code Management

https://github.com/JesusCrie/ModularBot

Download modularbot-nashorn-support

How to add to project

<!-- https://jarcasting.com/artifacts/com.jesus-crie/modularbot-nashorn-support/ -->
<dependency>
    <groupId>com.jesus-crie</groupId>
    <artifactId>modularbot-nashorn-support</artifactId>
    <version>2.5.0_23</version>
</dependency>
// https://jarcasting.com/artifacts/com.jesus-crie/modularbot-nashorn-support/
implementation 'com.jesus-crie:modularbot-nashorn-support:2.5.0_23'
// https://jarcasting.com/artifacts/com.jesus-crie/modularbot-nashorn-support/
implementation ("com.jesus-crie:modularbot-nashorn-support:2.5.0_23")
'com.jesus-crie:modularbot-nashorn-support:jar:2.5.0_23'
<dependency org="com.jesus-crie" name="modularbot-nashorn-support" rev="2.5.0_23">
  <artifact name="modularbot-nashorn-support" type="jar" />
</dependency>
@Grapes(
@Grab(group='com.jesus-crie', module='modularbot-nashorn-support', version='2.5.0_23')
)
libraryDependencies += "com.jesus-crie" % "modularbot-nashorn-support" % "2.5.0_23"
[com.jesus-crie/modularbot-nashorn-support "2.5.0_23"]

Dependencies

runtime (1)

Group / Artifact Type Version
com.jesus-crie : modularbot-core jar 2.5.0_23

Project Modules

There are no modules declared in this project.

ModularBot

Maven Central

There are probably typos and some language mistakes in this project, if you see one, notify me.

ModularBot is a framework aimed at anyone who would like to create a discord bot. It uses JDA to interface with discord; otherwise it provides a robust mechanism of modules that enables you to create small to large scale projects organized in feature modules and at the same time, provides you a organized and transparent way to manage them and their interactions.

It's modularity allows you to build only the modules you want to use instead of everything which means with the correct build configuration, you will be able to split your application in small chunks and leverage the amount of code that will be rebuilt and uploaded to your hosting service when you update your application.

The full framework comes in small modules that allows you to choose the exact features you want.

Side note: since v2.5.0, a module has been released that enables GraalVM users to write and build modules in every languages provided by GraalVM which means you are able to write modules in any languages with the robustness of JDA.

  1. Getting Started
  2. Modules
    1. Writing a Module
      1. Quickstart
      2. Lifecycle Hooks
      3. Injecting another Module
      4. Passing Parameters to your Module
      5. Registering/Requesting your Module
    2. Dependency Injection (DI)
      1. Requiring another Module (@InjectorTarget)
      2. Late Injections and Circular Dependencies (@LateInjectorTarget)
      3. Building Modules Manually
    3. Available Modules
      1. Core
      2. Console Logger
      3. Command
      4. Night Config Wrapper
      5. JS Nashorn Support
      6. JS Nashorn Command Support
      7. Message Decorator
      8. GraalVM Support
      9. GraalVM Support DiscordJS

Getting Started

The framework is fully available on Maven Central but you still need to add JCenter to your repositories to be able to download JDA dependencies.

repositories {
    mavenCentral()
    jcenter()
}

dependencies {
    implementation 'com.jesus-crie:modularbot-core:2.5.0_23'
    implementation 'com.jesus-crie:modularbot-logger:2.5.0_23'
    implementation 'com.jesus-crie:modularbot-command:2.5.0_23'
}

As every module of the framework has the same version you can define a global variable with the version as a shorthand to update every module at once.

Now you can build your first instance of ModularBot using the ModularBotBuilder class which mirrors the behaviour of the classic JDABuilder but with additional methods.

// Build a new instance of ModularBot with the base modules.
ModularBot bot = new ModularBotBuilder("token")
        .requestBaseModules() // Explicitly request every default modules
        .resolveAndBuild(); // If you use only #build(), the modules will not be instantiated !

// Register a quick command.
CommandModule module = bot.getModuleManager().getModule(CommandModule.class);
module.registerCreatorQuickCommand("stop", e -> bot.shutdown());

bot.login();

Note that this is a great way to start experiencing things but not the clean way to use the framework. Consider using modules as soon as possible and keep your main class clean.

Modules

Each module of your application is typically a collection of related utilities that provides the same kind of services, aka Feature Modules. Also each module is a singleton.

The usage of the term 'request' for saying that you want to use a module comes from the fact that you are actually asking the DI to provide you an instance of a given module.

All of the modules in this github repository are available on Maven Central.

Writing a module

To get a first glance of what a module is, it may be usefull the start writing one on your own to better understand how to manipulate them later.

Quickstart

Modules are just subclasses of Module and basically that's all that you need to create your first module.

public class MyModule extends Module {
    
    public MyModule() {
        // Hello world !
    }
}

Even the default constructor is optional here !

Note that just this isn't enough to tell the framework to create your module, we will register it in the last section of the chapter. Note also that you will almost never call any constructor directly, so don't make complex constructor with lots of overloads and all. Only one will be used.

Lifecycle hooks

Your modules can be notified of a few imoprtant steps of the lifecycle of the bot through the Lifecycle interface which defines some callbacks that you can use to do things at certain steps of the bootstrapping of your bot until its shutdown.

For example you can override the method Lifecycle#onLoad() to perform some initialization just after the module has been constructed and before the bot is created. Other hooks includes Lifecycle#onShardsReady(), triggered when the bot comes online and is ready to go.

The full listing of the callbacks is listed in the interface Lifecycle.

Injecting another module

You now have a module that can interact with the bot at any step of its lifecycle and thats great but you may want to split your code in multiple modules and communicate between them, or just use the default modules available like, for example the CommandModule.

There are two things required to inject a module:

  • First, you need to create a constructor that has the wanted module in its parameters.
  • Annotate this constructor with @InjectorTarget to tell the DI to use this constructor.

Important: You can only have one and only one annotated constructor.

For example if you want to write a module that register a command in the CommandModule, it will look like this:

public class MyModule extends Module {
    
    @InjectorTarget
    public MyModule(CommandModule cmdModule) {
        cmdModule.registerCreatorQuickCommand("ping", e -> e.fastReply("Pong !"));
    }
}

Note that if your module isn't explicitly requested nor injected in another module, it will never be instantiated by the DI.

Passing parameters to your module

You may want to make a configurable module and inject these settings in the constructor. You can define any other type of parameters in your constructor and pass arguments to them. Consider this module:

public class WelcomeModule extends Module {
    
    public WelcomeModule(NighConfigWrapperModule cfgModule, String messageFormat) {
        // ...
    }
    
    // ...
}

This module needs the config module to work, which will be provided automatically, but also need a string as its second parameter, which the framework can't guess.

You can define what will be the value passed this parameter in the ModularBotBuilder with ModularBotBuilder#configureModule():

ModularBot bot = new ModularBotBuilder("token")
    .requestModules(WelcomeModule.class)
    .configureModule(WelcomeModule.class, "Welcome %s !")
    .resolveAndBuild();

You can also specify default parameters in the module class by creating a static field of type ModuleSettingsProvider and annotating it with @DefaultInjectionParameters. These settings will be used if no call to #configureModule() has been made !

@DefaultInjectionParameters
private static ModuleSettingsProvider DEFAULT_SETTINGS = new ModuleSettingsProvider("Welcome %s !");

The order of your arguments matters, you need to provide them in the correct order but the presence of modules to be injected doesn't matter at all.

If your module has parameters and hasn't been configured in any way, or there were not enough arguments provided, they will be filled with the null value.

Registering/Requesting your module

If your module isn't registered implicitly (aka injected in another module), you need to do it explicitly in the builder like:

ModularBot bot = new ModularBotBuilder("token")
    .requestModules(
            MyAwesomeModule.class,
            AnotherModule.class
    )
    .resolveAndBuild();

Don't forget to resolve the modules using one of #resolveModules(), #resolveModulesSilently() or #resolveAndBuild().

Note that #requestBaseModules() can be used to request all of the default/official modules.

Dependency Injection (DI)

This dependency injector is a bit inspired of the way that Angular module injection is used: defining a constructor with the requested modules, declaring the module and boom you have everything you want.

This section will not talk of the internals of this feature but rather of its usage.

Requiring another module (@InjectorTarget)

Like you saw in the last section, you can request another module implicitly by requiring it. It means writing a constructor with the annotation and the module in question as a parameter.

You can require as many modules as you want in this constructor, in theory you can even request the module twice or more as long as they meet those requirements:

  • They need to extend Module of course, otherwise they will be treated like common parameters and not injections.
  • They need to be instantiable modules, you can't inject a module by only knowing one of its superclasses which isn't instantiable.

One could declare a target like that:

public class MyModule extends Module {
    @InjectorTarget
    public MyModule(CommandModule cmdModule, NightConfigWrapperModule cfgModule) {
        // TODO
    }
}

About the annotation (@InjectorTarget), there need to be exactly one constructor annotated by module, otherwise an exception will be thrown.

The only exception to that is the empty constructor, if the only constructor of your module is a constructor without any parameters or the default constructor, the annotation is optional.

Late injections and circular dependencies (@LateInjectorTarget)

If for some reason you have a circular dependency, the DI will throw an exception. A circular dependency is described by the fact that a module A requires module B that in some way also requires module A. It can be a simple loop, A -> B -> A -> ... or a larger loop like A -> C -> B -> A -> ...

If such a loop shows up, you may have a flaw in your architecture and you should consider this as a warning. If for some reasons you need it, you can still take advantage of the late injection mechanism.

Late injection takes place after every requested module has been instantiated. The DI will look for methods and fields annotated with @LateInjectorTarget and extract required modules from the method signature / the field type. Unlike the main injector target, parameters other than modules are considered an error and the whole method/field will be ignored.

If the target is valid and the module has already been built, it will be assigned to the field / passed to the method. The late injection will not build any additional modules !! A late injector method can contains multiple injections.

Back to the case of the circular dependency, if you have something like A -> B -> A, you can choose to late inject one of the two. For the example, A will be late injected into B.

You will end up with something like:

public class A extends Module {
    
    // Main target
    @InjectorTarget
    public A(B moduleB) {}
}

public class B extends Module {
    
    // Late inject via a field
    @LateInjectorTarget
    private A moduleA;
    
    // Main target
    @InjectorTarget
    public B() {}
    
    // Late inject via a method
    @LateInjectorTarget
    public void whatever(A moduleA) {}
}

For the sake of the example, both a field and a method late injector target are used for the same module. This can work but just don't.

Here, module A requires the module B as specified in its target which lead to the instantiation of module B. Module B doesn't require anything from the point of view of the main target so it can be instantiated without anything else, module A requirements are fulfilled and A can be instantiated.

After everything has been instantiated, the late injection comes in place and sees the field target, query the corresponding module (A) and set its value to it. Then it looks for method targets, find the method and fulfill its dependencies like for the field.

If you've been paying attention, you will notice something a little problematic with this solution. If We request only module B, module A will never be instantiated because it is never required in the main injector target and the late injection will just throw a warning when it will see that it hasn't the module A built. So for this to work, you need to request the module A, which requires module B.

You can also take advantage of this late injection mechanism to make optional dependencies.

Building modules manually

The DI allows you to provide already built modules if for some reason you want to provide you own instances of modules.

This is particularly useful for the logger module which need to be built to start receiving logs.

ModularBot bot = new ModularBotBuilder("token")
    .provideBuiltModules(
            new ConsoleLoggerModule()
    )
    .resolveAndBuild();

Available modules

ModularBot provides a few default modules that covers the primary needs of any discord bot such as config files, commands, ...

Core

Javadocs core

Artifact: com.jesus-crie:modularbot-core.

This is the core of the framework, its contains the main classes like ModularBot, Module and the DI. If you want only the basic features without any more advanced features you can use this artifact.

It doesn't have any default logger, like the rest of the framework it uses the slf4j logger which a custom implementation is provided in the logger module and works out of the box without any configuration.

Console Logger

Javadocs logger

Artifact: com.jesus-crie:modularbot-logger.

Provides an implementation of SLF4J.

Works out of the box without the need of any configuration. But you can still configure the message format and the log level via the static variables in the module class.

Loggers are typically constants in the class where they are declared.

A typical setup for a logger looks like:

public class MyClass {
    private static final Logger LOG = LoggerFactory.getLogger("MyClass");
    
    public void whatever() {
        LOG.info("I like trains");
    }
}

You can listen to logs by yourself by adding a listener using the static ModularLogger#addListener() method.

You can change the log level with

ConsoleLoggerModule.MIN_LEVEL = ModularLog.Level.INFO;

Command

Javadocs command

Artifact: com.jesus-crie:modularbot-command

This module provides a complete command system to craft commands that looks like !command arg1 arg2 "arg 3" --explicit-option arg -i -o "implicit options"

With this system each command need to have a dedicated class that extends Command. You can provide basic information about the command using the @CommandInfo annotation this class to avoid using a constructor.

Note that only the constructors Command#Command() and Command#Command(AccessLevel) uses the annotation.

Note that if you want to specify an AccessLevel you need to use a constructor.

The AccessLevel of a command is a set of prerequisites that a user need to satisfy before using a command. It contains a set of permissions that the user need to satisfy if the command is executed in a guild, plus some flags and the ID of an user if you want to authorize only one person. However you can still override the method AccessLevel#check(CommandEvent) and implement your own checks.

The way that AccessLevels are made is a bit crappy so expect changes.

Each command have a set of CommandPatterns that correspond to a certain manner to type a command. These patterns can be found automatically when the command class is instantiated* by looking at the methods in the class annotated with @RegisterPattern.

Note that if you want to take full advantage of this system you need to provide the argument -parameters to your compiler to be able to read the names of your method parameters.

With this system a command that have this syntax !command <@User> add <String> can be automatically registered by a method signature like this:

@RegisterPattern
protected void someMethod(CommandEvent event, Options options, User user, Void add, String string) {}

or:

@RegisterPattern(arguments = {"USER", "'add'", "STRING"})
protected void someMethod(CommandEvent event, Options options, User user, Void add, String string) {}

Note that the strings provided in the annotations (except for the second) are the names of constants in the Argument class. You can register your own class that contains such constants annotated with @RegisterArgument with the method Argument#registerArguments(Class).

There is a variety of possibility to make such methods, all of them can be found in this Test class.

*: It's planned to do this at compile-time but for now it happens basically when the command is registered so when the bot is waking up.

Each command can accept a certain set of Options provided in the constructor or in the @CommandInfo. These options are totally optional and to not appear in the CommandPatterns. These are added at the end of the command implicitly (-f) or explicitly (--force).

Explicit options need to be prefixed with -- and the long name of the option whereas implicit ones are prefixed by - and followed by one or more letters each representing the short name of an option. If they are followed by a string it will be considered as the argument of the option (or the last in implicit options).

Once parsed these options are accessible through the Options object provided along the arguments to the patterns. The argument of each option is also present.

Note that like the Arguments, all of the Options names are constants in the Option class and you can register your own constants with Option#registerOptions(Class).

Experimental: In the CommandModule you can set flags to the command processor to modify the behaviour of the algorithm but it's experimental and can lead to unexpected behaviour. This feature isn't a priority so if your're a volunteer you can fork this repo and send a pull request.

Finally, you can listen to the success or the failure of a command typed by a user by registering your own CommandListener with CommandModule#addListener.

Night Config Wrapper

Javadocs config

Artifact: com.jesus-crie:modularbot-night-config-wrapper

This module uses NightConfig 3.6.0 to load, parse and save config files. You will be forced to have a "primary" config file. You can also register named groups of secondary config files as well as singleton config groups (basically a named config).

Note that the default config file contains information that will be delivered to the CommandModule like the "creator_id" and a list of custom prefixes for guilds. Note that the creator id will only be loaded whereas the custom prefixes will be loaded and saved when the module is unloaded.

You can customize the path of the default config by configuring the module but if you want to use a completely different FileConfig you will need to instantiate it yourself and provide it as a built module.

You can use secondary config files like this:

public class MyModule extends Module {
    @InjectorTarget
    public MyModule(NightConfigWrapperModule cfgModule) {
        FileConfig cacheFile = cfgModule.registerSingletonSecondaryConfig("cache", "./cache.json");
    }
}

This module is entirely based on Night Config and I really recommend you to read its documentation.

JS Nashorn support

Javadocs nashorn

Artifact: com.jesus-crie:modularbot-nashorn-support

As described in the JEP 335, the Nashorn JavaScript engine has been deprecated in Java 11. Therefore this module is considered deprecated.

As a replacement, consider using GraalVM and the associated module.

This module allows you to load modules in JavaScript using the Nashorn Script Engine. It will consider each subdirectory in ./scripts/ (or the specified base folder) as a module and will try to load the main.js of each one (if it exists) and will wrap any object in the module top-level variable into a module and send lifecycle events to it.

A module in JavaScript looks like this:

function TestModule() {
    this.log = LoggerFactory.getLogger("TestModule");
    this.info = new ModuleInfo("TestModule", "Author", "http://example.com", "1.0", 1);
    
    this.onInitialization = function() {
        this.log.info("Module initialized");
    }
}

var module = new TestModule();

See this section for more information about the custom modules.

For each script, a header is added that imports some essential classes. You can found this header here. It can be overridden if there is a file called _header.js in the scripts folder.

JS Nashorn Command Support

Javadocs nashorn command

Artifact: com.jesus-crie:modularbot-nashorn-command-support

As an extension of the nashorn support module, this module too is considered deprecated.

An extension to the JS module that provide a way to use the command module in JavaScript.

This module let you define a #getCommands() that returns an array of JavaScriptCommand that will be wrapped into real command objects and registered. But because of my poor skills in JavaScript you can't use the annotation system and you need to register your patterns explicitly like in the example below. Regardless of that, all of the other features are available.

with (baseImports) {
    with (commandImports) {
        
        /* Module declaration here */
        
        function getCommands() {
            // Create a typed array
            var commands = new JavaScriptCommandArray(1);
            commands[0] = testJSCommand;
            return commands;
        }
    
        var testJSCommand = JavaScriptCommand.from({
            aliases: ["testjs"],
            description: "A demo command in JavaScript",
            shortDescription: "A demo command",
            accessLevel: AccessLevel.EVERYONE,
            options: [Option.FORCE],
            
            // Create the patterns by hand
            patterns: [
                new CommandPattern(
                    [
                        Argument.forString("add"),
                        Argument.STRING
                    ], function (event, args, options) {
                        event.fastReply("You wan to add: " + args[0]);
                    }
                ),
                new CommandPattern(function (event, args, options) {
                    if (options.has("force"))
                        event.fastReply("Hi, i'm force");
                    else event.fastReply("Hi");
                })
            ]
        });
    }
}

Note that this code comes in addition to the module declaration. If a script doesn't contains a module, its entirely ignored.

You can also extends JavaScriptCommand but for some reason Nashorn do not evaluate the arrays correctly and messes up everything, but feel free to experiment and send me a pull request.

For convenience you can add these imports to your custom header:

var JavaScriptCommandArray = Java.type("com.jesus_crie.modularbot_nashorn_command_support.JavaScriptCommand[]");

var commandImports = new JavaImporter(com.jesus_crie.modularbot_nashorn_command_support, 
    com.jesus_crie.modularbot_command, 
    com.jesus_crie.modularbot_command.processing);

Message Decorator

Javadocs message decorator

Artifact: com.jesus-crie:modularbot-message-decorator

Decorators are objects that can be bound to a specific message to extend their behaviour by listening to specific events regarding their message. This module is mainly made to allow a bunch of interactions using the message's reactions (emotes under the message).

Every decorator extends MessageDecorator which stores the bound message and its timeout. When a decorator is triggered, MessageDecorator#onTrigger is called and when it times out, it will call MessageDecorator#onTimeout which will call MessageDecorator#destroy in most implementations.

Certain decorators implements Cacheable which allows them to be saved in a cache file when the bot is down and reloaded when the bot wakes up. This caching is done automatically when the decorator is registered.

Note that all of the lambdas that you can provide are serializable and will be serialized and this means that if your lambda uses variables that aren't in the lambda's parameters, they will be serialized too and can lead to unexpected errors.

From there you can use the AutoDestroyMessageDecorator which allows you to delete the bound message automatically after a certain period of time or when the bot is shutting down.

The other decorators extends ReactionDecorator which allows interactions by the intermediate of message reactions. These reactions are wrapped in DecoratorButtons that also contains an action to perform when the button is triggered.

They are 2 kind of reaction decorator, permanent ones and dismissible ones.

In the dismissible ones you can find AlertReactionDecorator which acts a bit like the AutoDestroyMessageDecorator but you can delete it earlier by clicking a reaction under the message. ConfirmReactionDecorator acts like a yes/no dialog box for the user.

In the permanent decorators you can find the PollReactionDecorator which allows you to turn a message into a poll by providing the allowed "votes" to it, then you can query the votes at any times.

Querying the votes can be expensive if there are too many emotes. Consider querying them as less often as possible.

There is also the PanelReactionDecorator which, like the poll, allows you to set a bunch of reactions under the message. But the panel decorator is made too handle more complex operations for each buttons. You need to extend this class before using it and create a method per button that you want and annotate it with @RegisterPanelAction(...). More details can be found in the javadoc of the class.

Note: The example bot demonstrates a bunch of these decorators and their possibilities.

GraalVM Support

Javadocs graalvm suport

Artifact: com.jesus-crie:modularbot-graalvm-support

This module need to be built using the GraalVM JDK and to be run on the GraalVM JRE.

GraalVM is an implementation of the JVM (Java Virtual Machine) that allows you to create polyglot applications, aka allowing the usage of multiple languages and communication between them under the same runtime.

This module allows you to write a module in any language supported by GraalVM and interact with it from the Java code.

The examples that will be given are assuming a module written in Javascript.

You can declare a module by creating a class and exporting the class object using the Polyglot API of GraalVM.

class MyJSModule {
    constructor() {
        console.log("Hello from javascript !");
    }
}

Polyglot.export("class", MyJSModule);

You don't need to explicitly extend anything but the framework will act like you are extending Module like any other module and will call the corresponding lifecycle hooks with the same signature as declared in the Lifecycle interface.

If you haven't declared the callback methods, they will not be called.

In order for your module to work, you need to write a wrapper in Java to interact with it.

The most simple wrapper will look like that:

public class MyJSModule extends GraalModuleWrapper {
    @InjectorTarget
    public MyJSModule() {
        super(new File("my-js-module.js"));
    }
}

The GraalModuleWrapper class extends the Module class and provides you a few methods to handle interaction with the wrapped module. It already extends every lifecycle hook and propagate them to the module.

If your module has a public method that is supposed to be exposed to the rest of the application, you need to wrap this method and convert the arguments from Java to the client language.

This example demonstrates this pattern:

class MyJSModule {
    constructor(commandModule) {
        commandModule.registerQuickCommand('ping', e => e.fastReply('Pong !'));
    }
    
    signMessage(str) {
        return str + ' <3';
    }
    
    consumeAction(promise) {
        promise.then(res => console.log(res));
    }
}
public class MyJSModule extends GraalModuleWrapper {
    @InjectorTarget
    public MyJSModule(CommandModule cmdModule) {
        super(new File("my-js-module.js"), cmdModule);
    }
    
    public String signMessage(String str) {
        return safeInvoke("signMessage", str).asString();
    }
    
    public void consumeAction(Supplier<String> action) {
        safeInvoke("consumeAction", GUtils.createJSPromise(getContext(), new JSPromiseExecutorProxy() {
            @Override
            public void run() {
                resolve(action.get());
            }
        }));
    }
}

The GUtils utility class contains a few methods like this one that can wrap a Functional Interface into a JS Promise object.

If you are writing in TypeScript, you can use the @types declared in the subproject ModularBot-TS-Types which declares a bunch of types for both the Polyglot API and the JDA classes.

GraalVM Support DiscordJS

Javadocs graalvm suport discordjs

Artifact: com.jesus-crie:modularbot-graalvm-support-discordjs

This module isn't ready yet.

This module will wrap everything in the Discord.JS API to allow peoples who prefer the DJS way to do things in Javascript modules.

Versions

Version
2.5.0_23
2.5.0_22
2.4.1_21
2.4.0_20
2.3.4_19
2.3.4_18
2.3.3_17
2.3.2_16
2.3.2_15
2.3.1_13
2.3.0_10
2.2.0