BeanShell
Simple Java Scripting

version 1.3

Table of Contents


Introduction

This document is about BeanShell. BeanShell is a small, free, embeddable Java source interpreter with object scripting language features, written in Java. BeanShell executes standard Java statements and expressions but also extends Java into the scripting domain with common scripting language conventions and syntax. BeanShell is a natural scripting language for Java.

Scripting vs. Application Languages

Traditionally, the primary difference between a scripting language and a compiled language has been in its type system: the way in which you define and use data elements. You might be thinking that there is a more obvious difference here - that of "interpreted" code vs. compiled code. But the compiler in and of itself does not fundamentally change the way you work with a language. Nor does interpreting a language necessarily make it more useful for what we think of as "scripting". It is the type system of a language that makes it possible for a compiler to analyze the structure of an application for correctness. Without types, compilation is reduced to just a grammar check and an optimization for speed. From the developer's perspective, it is also the type system that characterizes the way in which we interact with the code.

Types are good. Without strongly type languages it would be very hard to write large scale systems and make any assertions about their correctness before they are run. But working with types imposes a burden on the developer. Types are labels and labeling things can be tedious. It can be especially tedious during certain kinds of development or special applications where it is flexibility and not program structure that is paramount. There are times where simplicity and ease of use is a more important criterion.

This is not just rationalization to cover some underlying laziness. Productivity affects what people do and more importantly do *not* do in the real world, much more than you might think. There is a lot of important software that exists in the world today only because the cost/benefit ratio in some developer's mind reached a certain threshold.

Unit testing - one of the foundations of writing good code - is a prime example. Unit tests for well written code are, in general, vitally important as a collective but almost insignificant individually. It's a "tragedy of the commons" that leads individual developers to repeatedly weigh the importance of writing another unit test with working on "real code". Give developers have a tool that makes it easy to perform a test with a line or two of code they will probably use it. If, moreover, it is also a tool that they enjoy using during their development process - that saves the time, they will be even more inclined to use it.

Customizability through scripting also opens the door to applications that are more powerful than the sum of their parts. When users can extend, enhance, and add to their applications they use them in new and unexpected ways.

Scripting is powerful.

Tearing Down the Barriers

Traditionally scripting languages have traded in the power of types for simplicity. Most scripting languages distill the type system to just one or a handful of types such as strings, numbers, or simple lists. This is sufficient for many kinds of scripting.

Many scripting languages operate in a loose, unstructured land - a place dominated by text and course-grained tools. As such these scripting languages have evolved sophisticated mechanisms for working with these simple types (regular expressions, pipes, etc.). As a result there has developed a casm between the scripting languages and the application languages created by the collapse of the type system in-between. The scripting languages have remained a separate species, isolated and speaking a different dialect from their brothers the application languages.

BeanShell is a new kind of scripting language. BeanShell begins with the standard Java language and bridges it into the scripting domain in a natural way, but allowing the developer to relaxing types where appropriate. It is possible to write BeanShell scripts that look exactly like Java method code. But it's also possible to write scripts that look more like a traditional scripting language, while still maintaining the framework of the Java syntax.

BeanShell emulates typed variables and parameters when they are used. This allows you to "seed" your code with strong types where appropriate. You can "shore up" repeatedly used methods as you work on them, migrating them closer to Java. Eventually you may find that you want to compile these methods and maintain them in standard Java. With BeanShell this is easy. BeanShell does not impose a syntactic boundary between your scripts and Java.

But the bridge to Java extends much deeper than simple code similarity. BeanShell is one of a new breed of scripting languages made possible by Java's advanced reflection capabilities. Since BeanShell can run in the same Java virtual machine as your application, you can freely work with real, live, Java objects - passing them into and out of your scripts. Combined with BeanShell's ability to implement Java interfaces, you can achieve seamless and simple integration of scripting into your Java applications. BeanShell does not impose a type boundary between your scripts and Java.

History

What seems like an eternity ago, back in the summer of 1993, I was working at Southwestern Bell Technology Resources and I was infatuated with the Tcl/Tk scripting language. On the advice of someone at Sun I also began playing around a bit with the Oak language written by James Gosling. Little did I know that within just a few years Oak, which would become Java, would not only spark a revolution, but that I would be writing one of the first books on the new Java language (Exploring Java, O'Reilly & Associates) and creating Java's first scripting language, BeanShell, drawing inspiration from Tcl.

BeanShell's first public release was not until 1997, but I had been poking at it in one form or another for some time before that. BeanShell as a language became practical when Sun added reflection to the Java language in version 1.1. After that, and after having seen its value in helping me create examples and snippets for the second edition of my book, I decided to try to polish it up and release it.

BeanShell has slowly, but steadily gained popularity since then. It has grown in fits and spurts as its contributor's time has allowed. But recently BeanShell has achieved a sort of critical mass. BeanShell is distributed with Emacs as part of the JDE and with Sun Microsystem's NetBeans / Forte for Java IDEs. BeanShell is also bundled by BEA with their Weblogic application server. We've had reports of BeanShell being used everywhere from the high energy physics laboratory CERN, to classrooms teaching programming to nine year olds. BeanShell is being used in everything from large financial applications all the way down to embedded systems floating in Buoys in the pacific ocean. I attribute this success to the power of the open source development model and owe many thanks to everyone who has contributed.

Conclusion

I believe that BeanShell is the simplest and most natural scripting language for Java because it is, foremost, Java. BeanShell draws on a rich history of scripting languages for its scripting syntax and uses it to very conservatively extend the Java language into this new domain. I hope that you have half as much fun using BeanShell as I have had working on it and I welcome all comments and suggestions.


Quick Start

Welcome to BeanShell. This is a crash course to get you going. We'll leave out many important options and details. Please see the rest of the user's guide for more information.

Download and Run BeanShell

Download the latest JAR file from http://www.beanshell.org and start up BeanShell either in the graphical desktop mode or on the command line.

If you just want to start playing around you may be able to launch the BeanShell desktop by simply double clicking on the BeanShell JAR file. More generally however you'll want to add the jar to your classpath so that you can work with your own classes and applications easily.

To do this you can either drop the BeanShell JAR file into your Java extensions folder or add it to your classpath. (Important: If you put BeanShell in the extensions folder and wish to use it with BSF applications like Jakarta Ant you must install the bsf.jar in the same location).

To install as an extension place the bsh.jar file in your 
$JAVA_HOME/jre/lib/ext folder.  (OSX users: place the bsh.jar in 
/Library/Java/Extensions or ~/Library/Java/Extensions for individual users.)

Or add BeanShell to your classpath like this:

unix:     export CLASSPATH=$CLASSPATH:bsh-xx.jar
windows:  set classpath %classpath%;bsh-xx.jar

Tip:
You can modify the classpath from within BeanShell using the addClassPath() and setClassPath() commands.

You can then run BeanShell in either a GUI or command line mode:

    java bsh.Console       // run the graphical desktop
or
    java bsh.Interpreter   // run as text-only on the command line
or
    java bsh.Interpreter filename [ args ] // run script file

It's also possible to call BeanShell from within your own Java applications, to reach it in a remote server mode for debugging, to use it as a servlet, or even in an applet. See "BeanShell Modes of Operation" for more details.

The BeanShell GUI

The BeanShell GUI desktop is meant to allow some experimentation with the features of BeanShell. It is not intended to be a replacement for a full featured IDE. Please check out the jEdit editor for an example of a full featured development environment based in part on BeanShell scripting capabilities.

Upon starting the BeanShell in GUI mode a console window will open. By right clicking on the desktop background you can open additional console windows and other tools such as a simple class browser.

Each console window runs a separate instance of the BeanShell interpreter. The graphical console supports basic command history, line editing, cut and paste, and even class and variable name completion. From the console you can open a simple editor window. In it you can write scripts and use the 'eval' option to evaluate the text in the attached console's workspace or a new workspace.

Java Statements and Expressions

BeanShell understands standard Java statements, expressions, and method declarations. Statements and expressions are all of the normal things that you'd say inside a Java method such as variable declarations and assignments, method calls, loops, and conditionals.

You can use these exactly as they would appear in Java, however in BeanShell you also have the option of working with "loosely typed" variables. That is, you can simply omit the types of variables that you use (both primitives and objects). BeanShell will only signal an error if you attempt to misuse the actual type of the variable.

Here are some examples:

foo = "Foo";    
four = (2 + 2)*2/2;
print( foo + " = " + four );  // print() is a BeanShell command

// Do a loop
for (i=0; i<5; i++)
    print(i);   

// Pop up a frame with a button in it
button = new JButton( "My Button" );
frame = new JFrame( "My Frame" );
frame.getContentPane().add( button, "Center" );
frame.pack();
frame.setVisible(true);

Useful BeanShell Commands

In the previous example we used a convenient "built-in" BeanShell command called print(), to display values. print() does pretty much the same thing as System.out.println() except that it insures that the output always goes to the command line. print() also displays some types of objects (such as arrays) more verbosely than Java would. Another related command is show(), which toggles on and off automatic display of the result of every line you type.

Here are a few other examples of BeanShell commands:

See the complete list of BeanShell Commands for more information.

Tip:
BeanShell commands are not really "built-in" but are simply BeanShell scripts that are automatically loaded from the classpath. You can add your own scripts to the classpath to extend the basic command set.

Scripted Methods

You can declare and use methods in BeanShell just as you would in a Java class.

int addTwoNumbers( int a, int b ) {
    return a + b;
}

sum = addTwoNumbers( 5, 7 );  // 12

Bsh methods may also allow dynamic (loose) argument and return types.

add( a, b ) {
    return a + b;
}

foo = add(1, 2);            // 3
foo = add("Oh", " baby");   // "Oh baby"

Implementing Interfaces

Note: implementing arbitrary interfaces requires BeanShell be run under a Java 1.3 or higher environment.

You can use the standard Java anonymous inner class syntax to implement an interface type with a script. For example:

ActionListener scriptedListener = new ActionListener() {
    actionPerformed( event ) { ... }
}

You don't have to script all of the methods of an interface. You can opt to script only those that you intend to call if you want to. The calling code will simply throw an exception if it tries to invoke a method that isn't defined. If you wish to override the behavior of a large number of methods - say to produce a "dummy" adapter for logging - you can implement a special method signature: invoke(name, args) in your scripted object. The invoke() method is called to handle any undefined method invocations:

ml = new MouseListener() {
    mousePressed( event ) { ... }
    // handle the rest
    invoke( name, args ) { print("Method: "+name+" invoked!");
}

Scripted Objects

In BeanShell, as in JavaScript and Perl, method "closures" allow you to create scripted objects. You can turn the results of a method call into an object reference by having the method return the special value this. You can then use the reference to refer to any variables set during the method call. Useful objects need methods of course, so in BeanShell scripted methods may also contain methods at any level. For example:

foo() {
    print("foo");
    x=5;

    bar() {
        print("bar");
    }

    return this;
}

myfoo = foo();    // prints "foo"
print( myfoo.x ); // prints "5"
myfoo.bar();      // prints "bar"

If this "closure" thing seems strange to don't worry. It's just an evolutionary step that languages acquired along the path to Objects. Please see the user's manual for a more thorough explanation.

Within your scripts, BeanShell scripted objects (i.e. any 'this' type reference like myFoo in the previous example) can automatically implement any Java interface type. When Java code calls methods on the interface the corresponding scripted methods will be invoked to handle them. BeanShell will automatically "cast" your scripted object when you attempt to pass it as an argument to a method that takes an interface type. For passing script references outside of BeanShell, you can perform an explicit cast where necessary. Please see the user manual for full details.

Calling BeanShell From Your Application

You can evaluate text and run scripts from within your application by creating an instance of the BeanShell interpreter and using the eval() or source() commands. You may pass in variable references to objects you wish to use in scripts via the set() method and retrieve results with the get() method.

import bsh.Interpreter;

Interpreter i = new Interpreter();  // Construct an interpreter
i.set("foo", 5);                    // Set variables
i.set("date", new Date() ); 

Date date = (Date)i.get("date");    // retrieve a variable

// Eval a statement and get the result
i.eval("bar = foo*10");             
System.out.println( i.get("bar") );

// Source an external script file
i.source("somefile.bsh");

Tip:
In the above example the Interpreter's eval() method also returned the value of bar as the result of the evaluation.

Conclusion

We hope this brief introduction gets you started. Please see the full user manual for more details. Please consult the mailing list archives for more useful information. http://www.beanshell.org/

Basic Syntax

BeanShell is, foremost, a Java interpreter. So you probably already know most of what you need to start scripting with BeanShell. This section describes specifically what portion of the Java language BeanShell interprets and how BeanShell extends it or "loosens" it to be more scripting-language-like.

Standard Java Syntax

In a BeanShell script (and on the command line) you can type normal Java statements and expressions and display the results. Statements and expressions are the kinds of things you normally find inside of a Java method: variable assignments, method calls, math expressions, for-loops, etc.

Here are some examples:

/*
    Standard Java syntax
*/

// Use a hashtable
Hashtable hashtable = new Hashtable();
Date date = new Date();
hashtable.put( "today", date );

// Print the current clock value
print( System.currentTimeMillis() );

// Loop
for (int i=0; i<5; i++)
    print(i);

// Pop up a frame with a button in it
JButton button = new JButton( "My Button" );
JFrame frame = new JFrame( "My Frame" );
frame.getContentPane().add( button, "Center" );
frame.pack();
frame.setVisible(true);

You can also define your own methods and use them just as you would inside a Java class. We'll get to that in a moment.

Loosely Typed Java Syntax

In the examples above, all of our variables have declared types. e.g. "JButton button". Beanshell will enforce these types, as you will see if you later try to assign something other than a JButton to the variable "button" (you will get an error message). However BeanShell also supports "loose" or dynamically typed variables. That is, you can refer to variables without declaring them first and without specifying any type. In this case BeanShell will do type checking where appropriate at runtime. So, for example, we could have left off the types in the above example and written all of the above as:

/*
    Loosely Typed Java syntax
*/

// Use a hashtable
hashtable = new Hashtable();
date = new Date();
hashtable.put( "today", date );

// Print the current clock value
print( System.currentTimeMillis() );

// Loop
for (i=0; i<5; i++)
    print(i);

// Pop up a frame with a button in it
button = new JButton( "My Button" );
frame = new JFrame( "My Frame" );
frame.getContentPane().add( button, "Center" );
frame.pack();
frame.setVisible(true);

This may not seem like it has saved us a great deal of work. But you will see the difference when you come to rely on scripting as part of your development and testing process; especially for in interactive use.

When a "loose" variable is used you are free to reassign it to another type of Java object later. Untyped BeanShell variables can also freely hold Java primitive values like int and boolean. Don't worry, BeanShell always knows the real types and only lets you use the values where appropriate. For primitive types this includes doing the correct numeric promotion that the real Java language would do when you use them in an expression.

Exception Handling

Exception handling using try/catch blocks works just as it does in Java. For example:

try {
    int i = 1/0;
} catch ( ArithmeticException e ) {
    print( e );
}

But you can loosely type your catch blocks if you wish:

try {
    ...
} catch ( e ) { 
    print( "caught exception: "+e );
}

Basic Scoping of Variables

Note:
As of BeanShell version 1.3 the default scoping of loosely typed variables was changed to be more consistent with Java. BeanShell still supports an alternate scoping used in earlier versions. This mode can be enabled for legacy code by setting the system property "localscoping" to true. See appendix "Local Scoping".

Variable scoping in BeanShell behaves, wherever possible, just like that in Java. Ordinary Java, however, does not offer "loose" variables (variables that can be used without being declared first). So we must define their behavior within BeanShell. We'll see in the next section that untyped variables - variables that are not declared and not assigned a value elsewhere - default to the local scope. This means that, in general, if you assign a value to a variable without first declaring it, you are creating a new local variable in the current scope.

Blocks

Blocks are statements between curly braces {}. In BeanShell, as in Java, blocks define a level of scope for typed variables: typed variables declared within a block are local to the block. Other assignments within the block occur, as always, wherever the variable was defined.

Untyped variables in BeanShell, however, are not constrained by blocks. Instead they act as if they were declared at the outer (enclosing) scope's level. With this in mind, BeanShell code looks just like Java code. In BeanShell if you declare a typed variable within a block it is local to the block. But if you use an untyped variable (which looks just like an ordinary assignment in Java) it behaves as an assignment to the enclosing scope.

This will make sense with a few examples:

// Arbitrary code block
{
    y = 2;      // Untyped variable assigned
    int x = 1;  // Typed variable assigned
}
print( y ); // 2
print( x ); // Error! x is undefined.

// Same with any block statement: if, while, try/catch, etc.
if ( true ) {
    y = 2;      // Untyped variable assigned
    int x = 1;  // Typed variable assigned
}
print( y ); // 2
print( x ); // Error! x is undefined.

Variables declared in the for-init area of a for-loop follow the same rules as part of the block:

for( int i=0; i<10; i++ ) {  // typed for-init variable
    j=42;
}
print( i ); // Error! 'i' is undefined.
print( j ); // 42

for( z=0; z<10; z++ ) { }   // untyped for-init variable
print( z ); // 10

Variable Modifiers

The standard Java variable modifiers may be used on typed variables: private / protected / public, final, transient, volatile, static. Only 'final' is currently implemented. The others are currently ignored.

Modifiers may not be applied to untyped variables.

Convenience Syntax

In BeanShell you may access JavaBean properties as if they were fields:

button = new java.awt.Button();
button.label = "my button";  // Equivalent to: b.setLabel("my button");
print( button.label );       // Equivalent to print( b.getLabel() );

JavaBean properties are simply pairs of "setter" and "getter" methods that adhere to a naming convention. In the above example BeanShell located a "setter" method with the name "setLabel()" and used it to assign the string value. It then found the method named getLabel() to retrieve the value.

Boolean properties may optionally use the syntax "is" for their "getter". e.g.

Float f = new Float(42f);
print( f.infinite );  // Equivalent to print( f.isInfinite() ); // false

If there is any ambiguity with an actual Java field name of the object (e.g. label in the above example) then the actual field name takes precedence. If you wish to avoid any ambiguity BeanShell provides an additional, uniform syntax for accessing both Java Bean properties and Hashtable or Map entries. You may use the "{}" curly brace construct with a String identifier as a qualifier on any variable of the appropriate type:

b = new java.awt.Button();
b{"label"} = "my button";  // Equivalent to: b.setLabel("my button");

h = new Hashtable();
h{"foo"} = "bar";          // Equivalent to: h.put("foo", "bar");

Where the java.util.Collections API is available, Maps are also supported.

Enhanced 'for' Loop

BeanShell supports the Java 1.5 style enhanced for-loop for iterating over collections and array types. (Note that you do not have to be running Java 1.5 to use this feature).

List foo = getSomeList();

for ( untypedElement : foo ) 
	print( untypedElement ); 

for ( Object typedElement: foo ) 
	print( typedElement );

int [] array = new int [] { 1, 2, 3 };

for( i : array ) 
	print(i);

for( char c : "a string" )
	print( c ); 

Supported iterable types include all the obvious things.

See also the BshIterator API which supports the ehanced for-loop and allows iteration over these types using the dynamically loaded BeanShell Collection manager.

Switch Statements

In BeanShell, the switch statement may be used not only with numeric types but with objects. For example, you may switch on Dates and Strings which are compared for equality with their equals() methods:

dateobj = new Date();

switch( dateobj ) 
{
	case newYears:
		break;
	case christmas:
		break;
	default:
}

Auto Boxing and Unboxing

"Boxing" and "Unboxing" are the terms used to describe automatically wrapping a primitive type in a wrapper class and unwrapping it as necessary. Boxing is a feature of Java (SDK1.5) and has been supported in BeanShell for many years.

BeanShell supports boxing and unboxing of primitive types. For example:

int i=5;
Integer iw = new Integer(5);
print( i * iw );  // 25

Vector v = new Vector();
v.put(1); 
int x = v.getFirstElement();

Importing Classes and Packages

In BeanShell as in Java, you can either refer to classes by their fully qualified names, or you can import one or more classes from a Java package.

// Standard Java
import javax.xml.parsers.*;
import mypackage.MyClass;

In BeanShell import statements may appear anywhere, even inside a method, not just at the top of a file. In the event of a conflict, later imports take precedence over earlier ones.

A somewhat experimental feature is the "super import". With it you may automatically import the entire classpath, like so:

import *;

The first time you do this BeanShell will map out your entire classpath; so this is primarily intended for interactive use. Note that importing every class in your classpath can be time consuming. It can also result in a lot of ambiguities. Currently BeanShell will report an error when resolving an an ambiguous import from mapping the entire classpath. You may disambiguate it by importing the class you intend.

Tip:
The BeanShell which() command will use the classpath mapping capability to tell you where exactly in your classpath a specified class is located:
bsh % which( java.lang.String );
Jar: file:/usr/java/j2sdk1.4.0/jre/lib/rt.jar

See "Class Path Management" for information about modifying the BeanShell classpath at run-time with the addClassPath() or setClassPath() commands.

Also see "BeanShell Commands" for information about importing new BeanShell commands from the classpath.

Default Imports

By default, common Java core and extension packages are imported for you. They are, in the order in which they are imported:

Two BeanShell package classes are also imported by default:

Finally, we should mention that BeanShell commands may be imported from the classpath. The default commands are imported in the following way:

importCommands("/bsh/commands");

We will discuss how to import your own commands in a later section.

Tip:
The classes java.awt.List and java.util.List are both imported by default. Because java.util.List is imported later, as part of the java.util package, it takes precedence. To access java.awt.List simply import it in, or the java.awt package again your script. Later imports take precedence.

Document Friendly Entities

BeanShell supports special overloaded text forms of all common operators to make it easier to embed BeanShell scripts inside other kinds of documents (e.g XML).

@gt>
@lt<
@lteq<=
@gteq>=
@or||
@and&&
@bitwise_and&
@bitwise_or|
@left_shift<<
@right_shift>>
@right_unsigned_shift>>>
@and_assign&=
@or_assign|=
@left_shift_assign<<=
@right_shift_assign>>=
@right_unsigned_shift_assign>>>=


Scripted Methods

You can define define methods in BeanShell, just as they would appear in Java:

int addTwoNumbers( int a, int b ) {
    return a + b;
}

And you can use them in your scripts just as you would any Java method or "built-in" BeanShell command:

sum = addTwoNumbers( 5, 7 );

Just as BeanShell variables may be dynamically typed, methods may have dynamic argument and return types. We could, for example, have declared our add() method above like so:

add( a, b ) {
    return a + b;
}

In this case, BeanShell would dynamically determine the types when the method is called and attempt to "do the right thing":

foo = add(1, 2);
print( foo ); // 3

foo = add("Oh", " baby");
print( foo ); // Oh baby

In the first case Java performed arithmetic addition on the integers 1 and 2. (By the way, if we had passed in numbers of other types BeanShell would have performed the appropriate numeric promotion and returned the correct Java primitive type.) In the second case BeanShell performed the usual string concatenation for String types and returned a String object. This example is a bit extreme, as there are no other overloaded operators like string concatenation in Java. But it serves to emphasize that BeanShell methods can work with loose types.

Methods with unspecified return types may return any type of object (as in the previous example). Alternatively they may also simply issue a "return;" without a value, in which case the effective type of the method is "void" (no type). In either case, the return statement is optional. If the method does not perform an explicit "return" statement and the return type is not explicitly set to void, the value of the last statement or expression in the method body becomes the return value (and must adhere to any declared return typing).

Method Modifiers and 'throws' Clauses

The standard Java modifiers may be applied to methods: private / protected / public, synchronized, final, native, abstract, and static.

The synchronized modifier is the only modifier currently implemented. The others are ignored. The 'throws' clause of methods is checked for valid class type names, but is not otherwise enforced.

Synchronized methods are synchronized on the object representing the method's common parent scope, so they behave like Java methods contained in a class. We will return to this topic after discussing scripted objects and "closures".

// foo() and bar() are synchronized as if they were in a common class
synchronized foo() { }
synchronized bar() { }

Scoping of Variables and Methods

As in Java, a method can refer to the values of variables and method names from the enclosing scope (in Java the "enclosing scope" would be a class). For example:

a = 1;
anotherMethod() { ... }

foo() {
    print( a );
	a = a+1;
	anotherMethod();
}

// invoke foo()
foo();      // prints 1
print( a ); // prints 2

Variables and methods are "inherited" from the parent scope in the usual way. In the example above there are just two levels of scope: the top or "global" scope and the scope of the method foo(). Later we'll talk about scripting objects in BeanShell and see that there can be arbitrary levels of scoping involved. But the rules will be the same.

As in Java, a typed variable is not visible outside the scope in which it is declared. So declaring a variable with a type is a way to limit its scope or make a local variable. In BeanShell using an untyped or "loosely" typed variable is also equivalent to declaring a local variable. That is, if you use a variable that has not been defined elsewhere, it defaults to the local scope:

a = 1;

foo() {
	a = a + 1; // a is defined in parent scope
	b = 3;     // undefined, defaults local scope
	int c = 4; // declared local scope
}

// invoke foo()
print( a ); // prints 2
print( b ); // ERROR! b undefined
print( c ); // ERROR! c undefined

In the above example the variable 'a' is declared in the global scope. When its value is read and assigned inside of foo() the global value of 'a' will be affected.

The variable 'b' is a usage of an untyped variable. Since 'b' has not been declared or assigned a value in any enclosing scope, it becomes a local variable 'b' in the scope of foo. The variable 'c' is explicitly declared (with a type) in the scope of foo() and is therefore, of course, local to foo().

Later we'll see that BeanShell allows arbitrary nesting of methods. If we were to declare another method inside of foo() it could see all of these variables (a, b, and c) as it is also in the scope of foo().

Scoping of Loosely Typed Variables

As in Java, declaring a variable with a type will always make it local. Even if the variable exists in the outer scope, it will be hidden by the local variable declaration. But what of loosely typed variables? As we've seen, untyped variable usage looks just like an ordinary Java assignment. What do we do if we want to make a local variable with the same name as a global one? One answer would be to resort to declaring the variable with a type. But if we wish to continue working with loosely typed variables in this case we have two options: We can explicitly declare a loosely typed variable with the BeanShell 'var' type. Or we can simply qualify our assignment with the 'this.' qualifier.

If you wish to, you can explicitly declare an untyped variable (making it local) using the special type 'var'. e.g.

foo() {
	var a = 1;
}
foo();
print( a ); // ERROR! a is undefined!

'var' is a magic type in BeanShell that represents a loose (untyped) variable. The default value of a variable declared with 'var' is null.

Alternately, you can use the scope modifier 'this' to explicitly qualify the variable assignment and make it local.

foo() {
	this.a = 1;
}
foo();
print( a ); // ERROR! a is undefined!

In this example we used the modifier 'this' to qualify an untyped variable's scope and make it local. We will explain 'this' and what it means in BeanShell scripted methods in the next section on Scripted Objects.

Scope Modifier: 'super'

Within a method, it is possible to explicitly qualify a variable or method reference with the identifier 'super' in order to refer to a variable or method defined in an enclosing scope (the scope in which the method is defined or "higher"). e.g.

int a = 42;

foo() {
    int a = 97;
    print( a );
    print( super.a );
}

foo();  // prints 97, 42

As in Java, the 'super' modifiers tells the scoping to begin its search for the variable or method in the parent scope. In the case above, the variable 'a' by default refers to the variable in the local scope. By qualifying 'a' with 'super' we can refer to the variable 'a' in the global scope (the "topmost" scope).

So, we've seen that 'super' can be used to refer to the method's parent context. We'll see in the next section how 'this' and 'super' are used in scripting Objects in BeanShell.


Scripted Objects

Many people who use BeanShell use it to write scripts that work with existing Java classes and APIs, or perform other kinds of dynamic activities for their own applications at run-time without the aid of a compiler. Often this means writing relatively unstructured code - for example, a sequence of method invocations or loops, all contained in a single script file or eval() statement. In the previous section we saw that BeanShell is also capable of scripting methods, just like Java. Creating methods and new BeanShell commands (which are just methods in their own files) is the natural progression of organizing your scripts into re-usable and maintainable components.

Beyond methods and structured programming lie, of course, objects and the full breadth of object oriented programming. In Java objects are the products of classes. While BeanShell is compatible with standard Java syntax for statements, expressions, and methods, you can't yet script new Java classes within BeanShell. Instead, BeanShell allows you to script objects as "method closures", similar to the way it is done in Perl 5.x, JavaScript, and other object-capable scripting languages. This style of scripting objects (which we'll describe momentarily) is simple and flows very naturally from the style of scripting methods. The syntax, as you'll see, is a straightforward extension of the standard Java concept of referring to an object with a 'this' reference.

Note:
In standard Java, a method inside of an object (an instance method) may refer to the enclosing object using the special variable 'this'. For example:
    // MyClass.java
    MyClass {
        Object getObject() {
            return this; // return a reference to our object
        }
    }
In the example above, the getObject() method of MyClass returns a reference to its own object instance (an instance of the MyClass object) using 'this'.

The 'this' reference

As in most languages, an executing method in BeanShell has its own "local" scope that holds argument (parameter) variables and locally declared variables. For example, in the following code segment any variables that we might use within the foo() method will normally only be visible within the scope of foo() and for the lifetime of one particular foo() method invocation:

// Define the foo() method:
foo() {
    int bar = 42;
    print( bar );
}   

// Invoke the foo() method:
foo();  // prints 42

print( bar ); // Error, bar is undefined here 

In the above, the bar variable is local to foo() and therefore not available outside of the method invocation - it is thrown away when the method exits, just like a standard Java local variable.

Now comes the twist - In BeanShell you have the option to "hang on" to the scope of a method invocation after exiting the method by referring to the special 'this' reference. As in Java, 'this' refers to the current object context. The only difference is that in this case the context is associated with the method and not a class instance.

By saving the 'this' reference after the method returns, you can continue to refer to variables defined within the method, using the standard Java "." notation:

foo() {
    int bar = 42;
    return this;
}

fooObject = foo();
print( fooObject.bar ); // prints 42!

In the above, the value returned by the foo() method (the 'this' reference) can be thought of as an instance of a "foo" object. Each foo() method invocation effectively creates a new object; foo() is now not just a method, but a kind of object constructor.

In the above case our foo object is not so much an object, but really more of a structure. It contains variables (bar) but no "behavior". The next twist that we'll introduce is that BeanShell methods are also allowed to contain other methods:

foo() {
    bar() {
        ...
    }
}

Scripted methods may define any number of nested methods in this way, to an arbitrary depth. The methods are "local" to the method invocation.

Statements and expressions within the enclosing BeanShell method can call their "local" methods just like any other method. (Locally declared methods override outer-more methods like local variables hide instance variables in Java.) The enclosed methods are not directly visible outside of their enclosing method. However, as you might expect, we can invoke them as we would on a Java object, through an appropriate object reference:

foo() {
    int a = 42;
    bar() {
        print("The bar is open!");
    }
    
    bar();
    return this;
}

// Construct the foo object
fooObject = foo();     // prints "the bar is open!"
// Print a variable of the foo object
print ( fooObject.a )  // 42
// Invoke a method on the foo object
fooObject.bar();       // prints "the bar is open!"

Methods declared inside block structures within methods behave just as if they were declared directly in the method. i.e. there are no block-local methods. For example:

foo() {

	bar() { }

	if ( true ) {
		bar2() { }
	}

	return this;
}

In the above example the methods bar() and bar2() are both defined within foo().

In the next section we'll return to the topic of variable scoping and go into more depth about how to work with scripted methods and objects.


Scope Modifiers

Now that we've seen how methods can be nested and treated as objects, we can revisit the topic of variable scope and scope modifiers.

'this', 'super', and 'global'

In the "Scripted Methods" section we described the use of 'super' to refer to a method's parent scope (the scope in which the method is defined). And in the previous section we talked about super's brother 'this', which refers to the current method's scope, allowing us to think of a method scope as an object. Now we can see how these concepts are related. Any method scope, and indeed the 'global' scope, can be thought as an object context. A scripted object can be thought of as encapsulated in a parent scope that determines its "environment" - its inherited variables and methods.

The references 'this', 'super', and 'global' are really the same kind of reference - references to BeanShell method contexts, which can be used as scripted objects. From here on We'll refer to 'this', 'super', 'global', and any other reference to a scripted object context in general as a 'this' type reference.

Note:
If you print a 'this' type reference you'll see what it refers to:
    BeanShell 1.3 - by Pat Niemeyer (pat@pat.net)
    bsh % print( this );
    'this' reference (XThis) to Bsh object: global
    bsh % foo() { print(this); print(super); }
    bsh % foo();
    'this' reference (XThis) to Bsh object: foo
    'this' reference (XThis) to Bsh object: global

The above note shows that the foo() method's 'this' reference is local (named 'foo') and that it's parent is the global scope; the same scope in which foo is defined.

'global'

The scope modifier 'global' allows you to always refer to the top-most scope. In the previous note you can see that the top level script context is called "global" and that it appears again as the 'super' of our foo() method. The global context is always the top scope of the script. It is the global namespace of the current interpreter. Referring to 'super' from the top scope simply returns the same 'global' again.

  global.foo = 42; 

Global variables are not special in any way. Their visibility derives simply from the fact that they are in the topmost scope. However, for those who do not like the idea of qualifying anything with "global". You can always use a more object oriented approach like the following.

// Create a top level object to hold some state
dataholder = object();

foo() {
    ...
    bar() {
        dataholder.value = 42;
    }

    bar();
    print( dataholder.value );
}
    

In the above example we used a global object to hold some state, rather than putting the 'value' variable directly in the global scope.

Tip:
In the above example we used the BeanShell object() command to create an "empty" BeanShell scripted object context in which to hold some data. The object() command is just a standard empty method named object() that returns 'this'. The variable 'dataholder' above is a 'this' type reference and has all of the properties of any other BeanShell object scope.

Synchronized Methods Revisited

Now that we have covered the meaning of 'this' and 'super' with respect to BeanShell methods we can define the meaning of the 'synchronized' modifier for BeanShell methods. Synchronized BeanShell methods behave as if they were in a common class by synchronizing on their common 'super' reference object. For example, in the four cases in the following example, synchronization occurs on the same Java object. That object is the 'this' type reference of the global scope (a Beanshell object of type bsh.This):

print( this ); // 'this' reference (XThis) to Bsh object: global

// The following cases all synchronize on the same lock
synchronized ( this ) { }     // synchronized block
synchronized int foo () { }   // synchronized method foo()
synchronized int bar () { }   // synchronized method bar()
int gee() {
	synchronized( super ) { }  // synchronized blockinside gee() 
}


Scripting Interfaces

One of the most powerful features of BeanShell is the ability to script Java interfaces. This feature allows you to write scripts that serve as event handlers, listeners, and components of other Java APIs. It also makes calling scripted components from within your applications easier because they can be made to look just like any other Java object.

Anonymous Inner-Class Style

One way to get a scripted component to implement a Java interface is by using the standard Java anonymous inner class syntax to construct a scripted object implementing the interface type. For example:

buttonHandler = new ActionListener() {
    actionPerformed( event ) { 
        print(event);
    }
};

button = new JButton();
button.addActionListener( buttonHandler );
frame(button);

In the above example we have created an object that implements the ActionListener interface and assigned it to a variable called buttonHandler. The buttonHandler object contains the scripted method actionPerformed(), which will be called to handle invocations of that method on the interface.

Note that in the example we registered our scripted ActionListener with a JButton using its addActionListener() method. The JButton is, of course, a standard Swing component written in Java. It has no knowledge that when it invokes the buttonHandler's actionPerformed() method it will actually be causing the BeanShell interpreter to run a script to evaluate the outcome.

To generalize beyond this example a bit - Scripted interfaces work by looking for scripted methods to implement the methods of the interface. A Java method invocation on a script that implements an interface causes BeanShell to look for a corresponding scripted method with a matching signature (name and argument types). BeanShell then invokes the method, passing along the arguments and passing back any return value. When BeanShell runs in the same Java VM as the rest of the code, you can freely pass "live" Java objects as arguments and return values, working with them dynamically in your scripts; the integration can be seamless.

See also the dragText example.

'this' references as Interface Types

The anonymous inner class style syntax which we just discussed allows you to explicitly create an object of a specified interface type, just as you would in Java. But BeanShell is more flexible than that. In fact, within your BeanShell scripts, any 'this' type script reference can automatically implement any interface type, as needed. This means that you can simply use a 'this' reference to your script or a scripted object anywhere that you would use the interface type. BeanShell will automatically "cast" it to the correct type and perform the method delegation for you.

For example, we could script an event handler for our button even more simply using just a global method, like this:

actionPerformed( event ) {
    print( event );
}

button = new JButton("Foo!");
button.addActionListener( this );
frame( button ); 

Here, instead of making a scripted object to hold our actionPerformed() method we have simply placed the method in the current context (the global scope) and told BeanShell to look there for the method.

Just as before, when ActionEvents are fired by the button, your actionPerformed() method will be invoked. The BeanShell 'this' reference to our script implements the interface and directs method invocations to the appropriately named method, if it exists.

Note:
If you want to have some fun, try entering the previous example interactively in a shell or on the command line. You'll see that you can then redefine actionPerformed() as often as you like by simply entering the method again. Each button press will find the current version in your shell. In a sense, you are working inside a dynamic Java object that you are creating and modifying as you type. Neat, huh? Be the Bean!

Of course, you don't have to define all of your interface methods globally. You can create references in any scope, as we discussed in "Scripting Objects". For example, the following code creates a scripted message button object which displays a message when its pushed. The scripted object holds its own actionPerformed() method, along with a variable to hold the Frame used for the GUI:

messageButton( message ) {
    JButton button = new JButton("Press Me");
    button.addActionListener( this );
    JFrame frame = frame( button );
 
    actionPerformed( e ) {
        print( message );
        frame.setVisible(false);
    }
}

messageButton("Hey you!");
messageButton("Another message...");

The above example creates two buttons, with separate messages. Each button prints its message when pushed and then dismisses itself. The buttons are created by separate calls to the messageButton() method, so each will have its own method context, separate local variables, and a separate instance of the ActionListener interface handler. Each registers itself (its own method context) as the ActionListener for its button, using its own 'this' reference.

In this example all of the "action" is contained in messageButton() method context. It serves as a scripted object that implements the interface and also holds some state, the frame variable, which is used to dismiss the GUI. More generally however, as we saw in the "Scripting Objects" section, we could have returned the 'this' reference to the caller, allowing it to work with our messageButton object in other ways.

Interface Types and Casting

It is legal, but not usually necessary to perform an explicit cast of a BeanShell scripted object to an interface type. For example:

actionPerformed( event ) {
    print( event );
}

button.addActionListener( 
    (ActionListener)this ); // added cast

In the above, the cast to ActionListener would have been done automatically by BeanShell when it tried to match the 'this' type argument to the signature of the addActionListener() method.

Doing the cast explicitly has the same effect, but takes a different route internally. With the cast, BeanShell creates the necessary adapter that implements the ActionListener interface first, at the time of the cast, and then later finds that the method is a perfect match.

What's the difference? Well, there are times where performing an explicit cast to control when the type is created may be important. Specifically, when you are passing references out of your script, to Java classes that don't immediately use them as their intended type. In our earlier discussion we said that automatic casting happens "within your BeanShell scripts". And in our examples so far BeanShell has always had the opportunity to arrange for the scripted object to become the correct type, before passing it on. But it is possible for you to pass a 'this' reference to a method that, for example, takes the type 'Object', in which case BeanShell would have no way of knowning what it was destined for later. You might do this, for example, if you were placing your scripted objects into a collection (Map or List) of some kind. In that case, you can control the process by performing an explicit cast to the desired type before the reference leaves your script.

Another case where you may have to perform a cast is where you are using BeanShell in an embedded application and returning a scripted object as the result of an eval() or a get() variable from the Interpreter class. There again is a case where BeanShell has no way of knowing the intended type within the script. By performing an explicit cast you can create the type before the reference leaves your script.

We'll discuss embedded applications of BeanShell in the "Embedding BeanShell" section a bit later, along with the Interpreter getInterface() method, which is another way of accomplishing this type of cast from outside a script.

"Dummy" Adapters and Incomplete Interfaces

It is common in Java to see "dummy" adapters created for interfaces that have more than one method. The job of a dummy adapter is to implement all of the methods of the interface with stubs (empty bodies), allowing the developer to extend the adapter and override just the methods of interest.

We hinted in our earlier discussion that BeanShell could handle scripted interfaces that implement only the subset of methods that are actually used and that is indeed the case. You are free in BeanShell to script only the interface methods that you expect to be called. The penalty for leaving out a method that is actually invoked is a special run-time exception: java.lang.reflect.UndeclaredThrowableException, which the caller will receive.

The UndeclaredThrowableException is an artifact of Java Proxy API that makes dynamic interfaces possible. It says that an interface threw a checked exception type that was not prescribed by the method signature. This is a situation that cannot normally happen in compiled Java. So the Java reflection API handles it by wrapping the checked exception in this special unchecked (RuntimeException) type in order to throw it. You can get the underlying error using the exception's getCause() method, which will, in this case, reveal the BeanShell EvalError exception, reporting that the scripted method of the correct signature was not found.

The invoke() Meta-Method

BeanShell provides a very simple short-hand mechanism for scripting interfaces with large numbers of methods. You can implement the special method invoke( name, args ) in any scripted context. The invoke() method will be called to handle the invocation of any method of the interface that is not defined. For example:

mouseHandler = new MouseListener() {
    mousePressed( event ) { 
        print("mouse button pressed");  
    }

    invoke( method, args ) { 
        print("Undefined method of MouseListener interface invoked:"
            + name +", with args: "+args
        );
    }
};

In the above example we have neglected to implement four of the five methods of the MouseListener interface. They will be handled by the invoke() method, which will simply print the name of the method and its arguments. However since mousePressed() is defined it will be called for the interface.

Here is a slightly more realistic example of where this comes in handy. Let's use the invoke() method to print the names of methods called via the ContentHandler interface of the Java SAX API, while parsing an XML document.

import javax.xml.parsers.*;
import org.xml.sax.InputSource;

factory = SAXParserFactory.newInstance();
saxParser = factory.newSAXParser();
parser = saxParser.getXMLReader();
parser.setContentHandler( this );

invoke( name, args ) {
    print( name );
}

parser.parse( new InputSource(bsh.args[0]) );

By running this script with the XML file as an argument, we can see which of the dozen or so methods of the SAX API are being exercised by the structure of the document, without having to write a stub for each of them.

Tip:
You can use the invoke( name, args ) meta-method directly in your own scope or in the global scope as well, in which case you can handle arbitrary "unknown" method invocations yourself, perhaps to implement your own "virtual" commands. Try typing this on the command line:
  invoke(name,args) { print("Command: "+name+" invoked!"); }
  noSuchMethod(); // prints "Command: noSuchMethod() invoked!"

Threads - Scripting Runnable

BeanShell 'this' type references can implement the standard java.lang.Runnable interface. So you can declare a "run()" method in your bsh objects and make it the target of a Thread:

foo() {
    run() {
        // do work...
    }
    return this;
}

foo = foo();
// Start two threads on foo.run()
new Thread( foo ).start();
new Thread( foo ).start();

BeanShell is thread-safe internally, so as long as your scripts do not explicitly do anything ordinarily non-thread safe (e.g. access shared variables or objects) you can write multi-threaded scripts.

Note:
You can use the bg() "background" command to run an external script in a separate thread. See bg().

Limitations

When running under JDK 1.3 or greater BeanShell can script any kind of Java interface. However when running under JDK 1.2 (or JDK1.1 + Swing) only the core AWT and Swing interfaces are available. To support those legacy cases a special extension of the 'this' reference implementation (the bsh.This class) is loaded which implements these interfaces along with Runnable, statically.


Special Variables and Values

In addition to the scope modifiers: 'this', 'super', 'global', BeanShell supports a number of pre-defined system variables, "magic" values, and methods.

Special Values

Note:
The choice of "bsh" for the root system object name was somewhat unfortunate because it conflicts with the current package name for BeanShell (also bsh). This means that if you wish to work with BeanShell classes explicitly from BeanShell scripts (e.g. bsh.Interpreter) you must first import them, e.g.:
    import bsh.Interpreter;
    i=new Interpreter();

Special Members of 'this' type References

'this' type references have several "magic" members:

These magic references are primarily used by BeanShell commands.

Undefined Variables

You can test to see if a variable is defined using the special value void. For example:

if ( foobar == void )
    // undefined

You can return a variable to the undefined state using the unset() command:

a == void;  // true
a=5;
unset("a"); // note the quotes
a == void;  // true

Setting the Command Prompt

Users may set the command line prompt string for use in interactive mode by setting the value of the variable bsh.prompt or by defining the scripted method (or command) getBshPrompt().

If the command or method getBshPrompt() is defined it will be called to get a string to display as the user prompt. For example, one could define the following method to place the current working directory into their command prompt:

	getBshPrompt() { return bsh.cwd + " % "; }

The default getBshPrompt() command returns the value of the variable bsh.prompt if it is defined or the string "bsh % " if not. If the getBshPrompt() method or command does not exist, throws an exception, or does not return a String, a default prompt of "bsh % " will be used.


BeanShell Commands

BeanShell commands appear to the user as pre-defined methods such as print(), load(), or save(). BeanShell Commands can be implemented as scripted methods or compiled Java classes which are dynamically loaded on demand from the classpath. We'll talk about adding your own commands in the next section "Adding BeanShell Commands".

Tip:
You can easily override any BeanShell command simply by defining the method yourself in your script. For example:

print( arg ) {
    System.out.println( "You printed: " + arg );
}

If you define the method in the global scope it will apply everywhere. If you define it local to a scripted object it will only apply in that object context.

Commands Overview

This is a high level overview of the BeanShell command set. You can find full documentation for all BeanShell commands in the "BeanShell Commands Documentation" section of this manual. See also the "BshDoc" section which covers javadoc style documentation of BeanShell script files.

Interpreter Modes

The following commands affect general modes of operation of the interpreter.
exit() Exit the interpreter. (Also Control-D).
show() Turn on "show" mode which prints the result of every evaluation that is not of void type.
setAccessibility() Turn on access to private and protected members of Java classes.
server() Launch the remote access mode, allowing remote access to the interpreter from a web browser or telnet client.
debug() Turns on debug mode. Note: this is very verbose, unstructured output and is primarily of interest to developers.
setStrictJava() Turn on "strict Java" mode which enforces Java compatibility by dissallowing loose types and undeclared variables.

Output

The following commands are used for output:
print(), error() Print output to standard out or standard error. print() always goes to the console, whereas System.out may or may not be captured by a GUI console or servlet.
frame() Display the AWT or Swing component in a Frame

Source and Evaluation

The following commands are used for evaluation or to run external scripts or applications:
eval() Evaluate a string as if it were typed in the current scope.
source(), sourceRelative() Read an external script file into the interpreter and evaluate it in the current scope
run(), bg() Run an external file in a subordinate interpreter or in a background thread in a subordinate interpreter.
exec() Run a native executable in the host OS

Utilities

The following commands are useful utilities:
javap() Print the methods and fields of an object, similar to the output of javap
which() Like the Unix 'which' command for executables. Map the classpath and determine the location of the specified class.
load(), save() load a serializable object from a file or save one to a file. Special handling is provided for certain objects.
object() Create an "empty" object context to hold variables; analogous to a Map.

Variables and Scope

The following commands affect the current scope:
clear()Clear all variables, methods and imports from the current scope.
unset()Remove a variable from the current scope. (Return it to the "undefined" state).
setNameSpace() Set the current namespace to a specified scope. Effectively bind the current scope to a new parent scope.

Classpath

The following commands manipulate or access the classpath:
addClassPath(), setClassPath(), getClassPath() Modify the BeanShell classpath.
reloadClasses() Reload a class or group of classes.
getClass() Load a class explicitly taking into account the BeanShell classpath.
getResource() Get a resource from the classpath.

Files and Directories

The following commands work with files, directories, and the working directory:
cd(), pwd(), dir(), rm(), mv(), cat() Unix Style file commands.
pathToFile() Translate a relative path to an absolute path taking into account the BeanShell current working directory.

Desktop and Class Browser

The following commands work with GUI tools:
classBrowser(), browseClass() Open a class browser window or browse a specific class or object.
desktop() Launch the BeanShell GUI desktop.
setNameCompletion() Turn on or off name completion in the GUI console.


Note:
The dir() command is written in Java; primarily as a demonstration of how to do this when desired.


Adding BeanShell Commands

BeanShell Commands are scripted methods or compiled Java classes which are dynamically loaded from the classpath to implement a method. All of the standard commands we discuss in this manual live in the BeanShell JAR file under the path /bsh/commands.

Adding to the set of "prefab" commands supplied with BeanShell is as easy as writing any other BeanShell methods. You simply have to place your script into a file named with the same name as the command and place the file in the classpath. You may then "import" the commands with the importCommands() method.

Command files can be placed anywhere in the BeanShell classpath. You can use even use the addClassPath() or setClassPath() commands to add new command directories or JARs containing commands to your script at any time.

Hello World

For example, let's make a helloWorld() command:

// File: helloWorld.bsh
helloWorld() { 
    print("Hello World!");
}

Place the command file helloWorld.bsh in a directory or JAR in your classpath and import it with the importCommands() command. You can either set the classpath externally for Java or inside of BeanShell with the addClassPath() command. For example, suppose we have placed the file in the path: /home/pat/mycommands/helloWorld.bsh. We could then do:

addClassPath("/home/pat");  // If it's not already in our classpath
importCommands("/mycommands");

We can now use helloWorld() just like any other BeanShell command.

helloWorld(); // prints "Hello World!"

importCommands() will accept either a "resource path" style path name or a Java package name. Either one is simply converted to a resource path or Java package name as required to load scripts or compiled BeanShell command classes. A relative path (e.g. "mycommands") is turned into an absolute path by prepending "/". You may import "loose" commands (like unpackaged classes) at the top of the classpath by importing "/".

If for example you have placed BeanShell commands along with your other classes in a Java package called com.xyz.utils in your classpath, you can import those commands with:

// equivalent
importCommands("com.xyz.utils");
importCommands("/com/xyz/utils");

Imported commands are scoped just like imported classes. So if you import commands in a method or object they are local to that scope.

Overloaded Commands

BeanShell command scripts can contain any number of overloaded forms of the command method, e.g.:

// File: helloWorld.bsh
helloWorld() { 
	print("Hello World!"); 
}
helloWorld( String msg ) { 
	print("Hello World: "+msg); 
}

BeanShell will select the appropriate method based on the usual rules for methods selection.

Compiled Commands

You can also implement BeanShell commands as compiled classes instead of scripts if you wish. Your class name must simply be the name of the command (matching case as well) and it must implement one or more static invoke() methods who's signatures match a pattern. The first two arguments of the invoke() method must be the bsh.Interpreter and bsh.CallStack objects that provide context to all BeanShell scripts. Then any number (possibly zero) of arguments, which are the arguments of the command may follow. BeanShell will select the appropriate method based on the usual rules for methods selection.

The dir() command is an example of a BeanShell command that is implemented in Java. Let's look at a snippet from it to see how it implements a pair of invoke() methods for the dir() and dir(path) commands.

/**
	Implement dir() command.
*/
public static void invoke( Interpreter env, CallStack callstack ) 
{
	String dir = ".";
	invoke( env, callstack, dir );
}

/**
	Implement dir( String directory ) command.
*/
public static void invoke( 
	Interpreter env, CallStack callstack, String dir ) 
{
...
}

User Defined Commands with invoke()

It is useful to note that the invoke() meta-method which we described in the section "Scripting Interfaces" can be used directly in scope as well as through an object reference and one could use this to load arbitrary commands or implement arbitrary behavior for commands (undefined method calls). For example:

invoke( String methodName, Object [] arguments ) { 
	print("You invoked the method: "+ methodName );
}

// invoke() will be called to handle noSuchMethod()
noSuchMethod("foo"); 

invoke() is called to handle any method invocations for undefined methods within its scope. In this case we have declared it at the global scope.

Commands Scope

Scripted BeanShell commands are loaded when no existing method matches the command name. When a command script is loaded it is sourced (evaluated) in the 'global' scope of the interpreter. This means that once the command is loaded the methods declared in the command script are then defined in the interpreter's global scope and subsequent calls to the command are simply handled by the those methods as any other scripted method.

Note:
Note that this means that currently scripted commands may only be loaded once and then they are effectively cached.

Getting the Caller Context

A useful feature of BeanShell for command writers is the 'this.caller' reference, which allows you to create side effects (set or modify variables) in the method caller's scope. For example:

fooSetter() {
    this.caller.foo=42;
}

The above command has the effect that after running it the variable 'foo' will be set in the caller's scope. e.g.:

fooSetter();
print( foo ); // 42

It may appear that we could simply have used the 'super' modifier to accomplish this and in this case it would have worked. However it would not have been correct in general because the 'super' of fooSetter() always points to the same location - the scope in which it was defined. We would like fooSetter() to set the variable in whatever scope it was called from.

To reiterate: The 'super' of a method is always the context in which the method was defined. But the caller may be any context in which the method is used. In the following example, the parent context of foo() and the caller context of foo() are the same:

foo() { ... }
foo();

But this is not always the case, as for bar() in the following example:

foo() { 
    bar() { ... }
    ...
}

// somewhere
fooObject.bar();

The special "magic" field reference: 'this.caller' makes it possible to reach the context of whomever called bar(). The 'this.caller' reference always refers to the calling context of the current method context.

The diagram above shows the foo() and bar() scopes, along with the caller's scope access via 'this.caller'.

This is very useful in writing BeanShell commands. BeanShell command methods are always loaded into the global scope. If you refer to 'super' from your command you will simply get 'global'. Often it is desirable to write commands that explicitly have side effects in the caller's scope. The ability to do so makes it possible to write new kinds of commands that have the appearance of being "built-in" to the language.

A good example of this is the eval() BeanShell command. eval() evaluates a string as if it were typed in the current context. To do this, it sends the string to an instance of the BeanShell interpreter. But when it does so it tells the interpreter to evaluate the string in a specific namespace: the namespace of the caller; using this.caller.

    eval("a=5");
    print( a ); // 5

The eval() command is implemented simply as:

eval( String text ) {
	this.interpreter.eval( text, this.caller.namespace );
}

As a novelty, you can follow the call chain further back if you want to by chaining the '.caller' reference, like so:

    this.caller.caller...;

Or, more generally, another magic reference 'this.callstack' returns an array of bsh.NameSpace objects representing the full call "stack". This is an advanced topic for developers that we'll discuss in another location.

setNameSpace()

In the previous discussion we used the this.caller reference to allow us to write commands that have side effects in the caller's context. This is a powerful tool. But what happens when one command calls another command that intends to do this? That would leave the side effects in the first command's context, not it's original caller. Fortunately this doesn't come up all that often. But there is a general way to solve this problem. That is to use the powerful setNameSpace() method to "step into" the caller's context. After that we may set variables and call methods exactly as if we were in the caller's context (because we are). If all commands did this there would be no need to use the this.caller reference explicitly (indeed, we may make it idiomatic for all commands to do this in the future).

myCommand() {
	// "Step into" the caller's namespace.
	setNameSpace( this.caller.namespace );

	// work as if we were in the caller's namespace.
}

You can try out the setNameSpace() command with arbitrary object scope's as well. For example:

object = object();

// save our namespace
savedNameSpace = this.namespace;

// step into object's namespace
setNameSpace( object.namespace );

// Work in the object's scope
a=1;
b=2;

// step back 
setNameSpace( savedNameSpace );

print( object.a ); // 1
print( object.b ); // 2

print( a ); // ERROR! undefined

Getting the Invocation Text

You can get specific information about the invocation of a method using namespace.getInvocationLine() and namespace.getInvocationText(). The most important use for this is in support of the ability to write an assert() method for unit tests that automatically prints the assertion text.

assert( boolean condition ) 
{
    if ( condition )
        print( "Test Passed..." );
    else {
        print(
            "Test FAILED: "
            +"Line: "+ this.namespace.getInvocationLine()
            +" : "+this.namespace.getInvocationText()
            +" : while evaluating file: "+getSourceFileInfo()
        );
        super.test_failed = true;
    }
}

Working with Dirctories and Paths

BeanShell supports the notion of a current working directory for commands that work with files. The cd() command can be used to change the working directory and pwd() can be used to display the current value. The BeanShell current working directory is stored in the variable bsh.cwd.

All commands that work with files respect the working directory, including the following:

pathToFile()

As a convenience for writing your own scripts and commands you can use the pathToFile() command to translate a relative file path to an absolute one relative to the current working directory. Absolute paths are unmodified.

absfilename = pathToFile( filename );

Path Names and Slashes

When working with path names you can generally just use forward slashes in BeanShell. Java localizes forward slashes to the appropriate value under Windows environments. If you must use backslashes remember to escape them by doubling them:

dir("c:/Windows"); // ok
dir("c:\\Windows"); // ok

Working With Class Identifiers

You may have noticed that certain BeanShell commands such as javap(), which(), and browseClass() which take a class as an argument can accept any type of argument, including a plain Java class identifier. For example, all of the following are legal:

javap( Date.class ); // use a class type directly
javap( new Date() ); // uses class of object
javap( "java.util.Date" ); // Uses string name of class
javap( java.util.Date );  // Use plain class identifier

In the last case above we used the plain Java class identifier java.util.Date. In Beanshell this resolves to a bsh.ClassIdentifier reference. You can get the class represented by a ClassIdentifier using the Name.identifierToClass() method. Here is an example of how to work with all of the above, converting the argument to a class type:

	import bsh.ClassIdentifier;

    if ( o instanceof ClassIdentifier )
        clas = this.namespace.identifierToClass(o);
    if ( o instanceof String)
        clas = this.namespace.getClass((String)o);
    else if ( o instanceof Class )
        clas = o;
    else
        clas = o.getClass();

Working with Iterable Types

In conjunction with the enhanced for-loop added in BeanShell version 1.3 a unified API was added to provide support for iteration over composite types. The bsh.BshIterator interface provides the standard hasNext() and next() methods of the java.util.Iterator interface, but is available in all versions of Java and can be created for all composite types including arrays.

The BeanShell CollectionManager is used to get a BshIterator for an interable object or array. It is a dynamically loaded extension, so it provides support for the java.util.Collections API when available, but does not break compatability for Java 1.1 applications. You can use this in the implementation of BeanShell commands to iterate over Enumeration, arrays, Vector, String, StringBuffer and (when the java.util.collections API is present) Collections and Iterator.

cm = CollectionManager.getCollectionManager();
if ( cm.isBshIterable( myObject ) ) 
{
	BshIterator iterator = cm.getBshIterator( myObject );
	while ( iterator.hasNext() )
		i = iterator.next();
}


Strict Java Mode

Note: Strict Java Mode is new and currently breaks some BeanShell tools and APIs when activated. The GUI desktop and most BeanShell commands will not work with strict Java mode enabled. Please see notes at the end of this page

If you are a Java teacher or a student learning the Java language and you would like to avoid any potential confusion relating to BeanShell's use of loose variable types, you can turn on Strict Java Mode. Strict Java Mode is enabled with the the setStrictJava() command. When strict Java mode is enabled BeanShell will require typed variable declarations, method arguments and return types. For example:

setStrictJava(true);

int a = 5;

foo=42; // Error! Undeclared variable 'foo'.

bar() { .. } // Error! No declared return type.


Class Loading and Class Path Management

BeanShell is capable of some very fine grained and sophisticated class reloading and modifications to the class path. BeanShell can even map the entire class path to allow for automatic importing of classes.

Changing the Class Path

addClassPath( URL | path )

Add the specified directory or archive to the classpath. Archives may be located by URL, allowing them to be loaded over the network.

Examples:

addClassPath( "/home/pat/java/classes" );
addClassPath( "/home/pat/java/mystuff.jar" );
addClassPath( new URL("http://myserver/~pat/somebeans.jar") );

Note that if you add class path that overlaps with the existing Java user classpath then the new path will effectively reload the classes in that area.

If you add a relative path to the classpath it is evaluated to an absolute path; it does not "move with you".

cd("/tmp");
addClassPath("."); // /tmp

setClassPath( URL [] )

Change the entire classpath to the specified array of directories and/or archives.

This command has some important side effects. It effectively causes all classes to be reloaded (including any in the Java user class path at startup). Please see "Class Reloading" below for further details.

Note: setClassPath() cannot currently be used to make the classpath smaller than the Java user path at startup.

Auto-Importing from the Classpath

As an alternative to explicitly importing class names you may use the following statement to trigger automatic importing:

import *;

There may be a significant delay while the class path is mapped. This is why auto-importing is not turned on by default. When run interactively, Bsh will report the areas that it is mapping.

It is only necessary to issue the auto-import command once. Thereafter changes in the classpath via the addClassPath() and setClassPath() commands will remap as necessary.

Note: As of BeanShell 1.1alpha new class files added to the classpath (from outside of BeanShell) after mapping will not be seen in imports.

Reloading Classes

BeanShell provides an easy to use mechanism for reloading classes from the classpath. It is possible in BeanShell to reload arbitrary subsets of classes down to a single class file. However There are subtle issues to be understood with respect to what it means to reload a class in the Java environment. Please see the discussion of class loading detail below. But in a nutshell, it is important that classes which work together be reloaded together at the same time, unless you know what you are doing.

reloadClasses( [ package name ] )

The most course level of class reloading is accomplished by issuing the reloadClasses() command with no arguments.

reloadClasses();

This will effectively reload all classes in the current classpath (including any changes you have made through addClassPath()).

Note: that reloading the full path is actually a light weight operation that simply replaces the class loader - normal style class loading is done as classes are subsequently referenced.

Be aware that any object instances which you have previously created may not function with new objects created by the new class loader. Please see the discussion of class loading details below.

You can also reload all of the classes in a specified package:

reloadClasses("mypackage.*");

This will reload only the classes in the specified package. The classes will be reloaded even if they are located in different places in the classpath (e.g. if you have some of the package in one directory and some in another).

As a special case for reloading unpackaged classes the following commands are equivalent:

reloadClasses(".*") 
reloadClasses("<unpackaged>")

You can also reload just an individual class file:

reloadClasses("mypackage.MyClass") 

Note: As of alpha1.1 classes contained in archives (jar files) cannot be reloaded. i.e. jar files cannot be swapped.

Mapping the path

Unlike the reloadClases() command which reloads the entire class path, when you issue a command to reload a package or individual class name BeanShell must map some portions of the classpath to find the location of those class files. This operation can be time consuming, but it is only done once. If running in interactive mode feedback will be given on the progress of the mapping.

Loading Classes Explicitly

In order to perform an explicit class lookup by name while taking into account any BeanShell class path modification you must use a replacement for the standard Class.forName() method.

The getClass() command will load a class by name, using the BeanShell classpath. Alternately, you can consult the class manager explicitly:

name="foo.bar.MyClass";
c = getClass( name );
c = BshClassManager.classForName( name );  // equivalent

Setting the Default ClassLoader

The bsh.Interpeter setClassLoader() and bsh.BshClassManager.setClassLoader() methods can be used to set an external class loader which is consulted for all basic class loading in BeanShell.

BeanShell will use the specified class loader at the same point where it would otherwise use the plain Class.forName(). If no explicit classpath management is done from the script (addClassPath(), setClassPath(), reloadClasses()) then BeanShell will only use the supplied classloader. If additional classpath management is done then BeanShell will perform that in addition to the supplied external classloader. However BeanShell is not currently able to reload classes supplied through the external classloader.

Class Loading in Java

A fundamental Java security proposition is that classes may only be loaded through a class loader once and that classes loaded through different class loaders live in different name spaces. By different name spaces I mean that they are not considered to be of the same type, even if they came from the very same class file.

You can think of this in the following way: When you load classes through a new class loader imagine that every class name is prefixed with the identifier "FromClassLoaderXXX" and that all internal references to other classes loaded through that class loader are similarly rewritten. Now if you attempt to pass a reference to a class instance loaded through another class loader to one of your newly loaded objects, it will not recognize it as the same type of class.

BeanShell works with objects dynamically through the reflection API, so your scripts will not have a problem recognizing reloaded class objects. However any objects which have you already created might not like them.

Class Loading in BeanShell

The following is a discussion of the BeanShell class loader architecture, which allows both course class path extension and fine grained individual class reloading.

Thriftiness - Abiding by the BeanShell thriftiness proposition: no class loading code is exercised unless directed by a command. BeanShell begins with no class loader and only adds class loading in layers as necessary to achieve desired effects.

The following diagram illustrates the two layer class loading scheme:

A "base" class loader is used to handle course changes to the classpath including added path. Unless directed by setClassPath() the base loader will only add path and will not cover existing Java user class path. This prevents unnecessary class space changes for the existing classes.

Packages of classes and individual classes are mapped in sets by class loaders capable of handling discrete files. A mapping of reloaded classes is maintained. The discrete file class loaders will also use this mapping to resolve names outside there space, so when any individual class is reloaded it will see all previously reloaded classes as well.

The BshClassManager knows about all class loader changes and broadcasts notification of changes to registered listeners. BeanShell namespaces use this mechanism to dereference cached type information, however they do not remove existing object instances.

Type caching is extremely important to BeanShell performance. So changing the classloader, which necessitates clearing all type caches, should be considered an expensive operation.


Modes of Operation

There are currently five basic modes of operation for running BeanShell: We'll outline these in this and the coming sections.

BeanShell is also integrated into a number of other tools and development environments including the Emacs JDE and the NetBeans/Forte for Java IDE. Please see the web site for articles and information about using BeanShell within these third party tools.

Standalone


You can use BeanShell to run scripts from the command line or enter statements interactively by starting the bsh.Interpreter class. (See "Quickstart" for instructions on adding BeanShell to your classpath.)

 java bsh.Interpreter [ filename ] [ arg ] [ ... ]  // Run a script file

There are a few options which can be passed to the Interpreter using Java system properties:

Remote

The bsh.Remote launcher is the equivalent of bsh.Interpreter, but runs the specified file in a remote BeanShell engine. The remote engine may be a servlet mode BeanShell engine (BshServlet) or a native server mode remote BeanShell instance (embedded interpreter).

bsh.Remote accepts a URL and filename as arguments:

// servlet mode URL
java bsh.Remote http://localhost/bshservlet/eval test1.bsh

// remote server mode URL
java bsh.Remote bsh://localhost:1234/ test1.bsh

An HTTP URL may be specified that points to an instance of BshServlet (See "Servlet Mode" for details). Or a native "bsh:" URL may be specified, pointing to an instance of the BeanShell interpreter running in remote server mode. At the time of this writing bsh: style URLs for accessing native remote server mode instances are not implemented.

In either case, bsh.Remote sends the script to the remote engine for evaluation. If Remote can parse the retun value of the script as an integer it will return the value as the exit status to the command line.

Interactive Use

One of the most popular uses for BeanShell is, of course, as a "shell" for interactive experimentation and debugging. BeanShell can be run in a GUI desktop mode that offers a number conveniences like command line history, cut & paste, and tools for interactive use such as a simple classbrowser. We'll talk about the GUI in "The BeanShell Desktop" later.

Tip:
The BeanShell GUI is comprised mostly of a set of BeanShell scripts supplied in the JAR file and launched by the BeanShell desktop() command.

However BeanShell can also be run interactively in plain text on the command line.

 java bsh.Interpreter   // Run interactively on the command line

This is useful for quick "one liners"; however it does not offer creature comforts such as command line history and editing. We should note that some shells, such as the Windows environment, do command line history and editing automatically - providing these features for BeanShell.

Tip:
You can exit from an interactive shell by typing Control-D.

The return statement is ignored in interactive mode (it does not exit the shell).

The .bshrc Init File

When run interactively, BeanShell looks for a startup file called ".bshrc" in the user's home directory. If the file is found it is sourced into the interactive shell. You can use this to perform setup for your interactive use. For example, to add additional default imports or to toggle on the show() command if you prefer that.

The location of the .bshrc file is determined by the Java system property "user.home", which has different locations under different operating systems. The following table lists common locations:

.bshrc File Location:

Unix$HOME/.bshrc
Win95/98 single user C:\Windows\.bshrc
Win98 Multiuser C:\Windows\Profiles\<username>\.bshrc
NT/2K C:\Winnt\Profiles\<username>\.bshrc

Embedding BeanShell in Your Application


BeanShell was designed not only as a standalone scripting language - to run scripts from files or on the command line - but to be easily embeddable in and extensible by your applications. When we talk about embedding BeanShell in your application we mean simply that you can use the BeanShell Interpreter in your own classes, to evaluate scripts and work with objects dynamically.

There are a number of reasons you might use BeanShell in this way. Here are a few:

Highly Customizable Applications

You can use BeanShell to make your applications highly customizable by users without requiring them to compile Java classes or even to know all of the Java syntax. During development you can use BeanShell to "factor out" volatile or environmentally dependent algorithms from your application and leave them in the scripting domain while they are under the most intense change. Later it is easy to move them back into compiled Java if you wish because BeanShell syntax is Java syntax.

Macros and Evaluation

BeanShell can be used as a generic language for "macros" or other complex tasks that must be described by a user of your application. For example, the popular JEdit Java editor uses BeanShell to allow users to implement macros for key bindings. This gives user power to customize the behavior of the editor, using as much (or as little) of the full power of Java as desired.

Java with loose variables is a very simple and appealing language; especially because there is already so much Java out there. Even a non-programmer will be familiar with the name "Java" and more inclined to want to work with it than an arbitrary new language.

BeanShell can also be used to perform dynamic evaluation of complex expressions such as mathematics or computations entered by the user. Why write an arithmetic expression parser when you can let your user enter equations using intermediate variables, operators, and other constructs. If strict control is desired, you can generate the script yourself using your own rules, and still leave the evaluation to BeanShell.

Simplify Complex Configuration Files

Many applications use simple Java properties files or XML for the majority of their runtime configuration. It is very common in the development of a large applications for configuration files like this to become increasingly complex. It can begin in a number of seemingly harmless ways - with the desire to make "cross references" inside the config files (XML supports this nicely). Then comes the desire to do something like variable substitution - which introduces some new syntax such as "${variable}" and usually a second pass in the parsing stage. Usually, at some point, integration with Java forces the introduction of class names into the mix. The configuration files soon want to start assigning parameters for object construction. Ultimately what you'll discover is that you are creating your own scripting language - and one that is probably not as easy to read as plain old Java.

BeanShell solves the problem of complex configuration files by allowing users to work not only with simple properties style values (loose variable assignment) but also to have the full power of Java to construct objects, arrays, perform loops and conditionals, etc. And as we'll see, BeanShell scripts can work seamlessly with objects from the application, without the need to turn them into strings to cross the script boundary.

The BeanShell Core Distribution

Beanshell is fairly small, even in its most general distribution. The full JAR with all utilities weighs in at about 250K. But BeanShell is also distributed in a componentized fashion, allowing you to choose to add only the utilities and other pieces that you need. The core distribution includes only the BeanShell interpreter and is currently about 130K. We expect this size to drop in the future with improvements in the parser generator. Any significant new features will always be provided in the form of add-on modules, so that the core language can remain small.

More and more people are using BeanShell for embedded applications in small devices. We have reports of BeanShell running everywhere from p