Overview

Serp is an open source framework for manipulating Java bytecode. If you have a suggestion for functionality that you would like to see incorporated into serp, please feel free to send it to me. Bug reports are also greatly appreciated. And if you choose to incorporate serp into any of your projects, I would love to hear about it!

Abe White
awhite@bea.com

Contents

1. Bytecode
2. Project Information
3. Javadoc
4. Download
5. License

Bytecode

Bytecode manipulation is a powerful tool in the arsenal of the Java developer. It can be used for tasks from compiling alternative programming languages to run in a JVM, to creating new classes on the fly at runtime, to instrumenting classes for performance analysis, to debugging, to altering or enhancing the capabilities of existing compiled classes. Traditionally, however, this power has come at a price: modifying bytecode has required an in-depth knowledge of the class file structure and has necessitated very low-level programming techniques. These costs have proven too much for most developers, and bytecode manipulation has been largely ignored by the mainstream.

The goal of the serp bytecode framework is to tap the full power of bytecode modification while lowering its associated costs. The framework provides a set of high-level APIs for manipulating all aspects of bytecode, from large-scale structures like class member fields to the individual instructions that comprise the code of methods. While in order to perform any advanced manipulation, some understanding of the class file format and especially of the JVM instruction set is necessary, the framework makes it as easy as possible to enter the world of bytecode development.

There are several other excellent bytecode frameworks available. Serp excels, however, in the following areas:

• Ease of use. Serp provides very high-level APIs for all normal bytecode modification functionality. Additionally, the framework contains a large set of convenience methods to make code that uses it as clean as possible. From overloading its methods to prevent you from having to make type conversions, to making shortcuts for operations like adding default constructors, serp tries to take the pain out of bytecode development.
• Power. Serp does not hide any of the power of bytecode manipulation behind a limited set of high-level functions. In addition to its available high-level APIs, which themselves cover the functionality all but the most advanced users will ever need, serp gives you direct access to the low-level details of the class file and constant pool. You can even switch back and forth between low-level and high-level operations; serp maintains complete consistency of the class structure at all times. A change to a method descriptor in the constant pool, for example, will immediately change the return values of all the high-level APIs that describe that method.
• Constant pool management. In the class file format, all constant values are stored in a constant pool of shared entries to minimize the size of class structures. Serp gives you access to the constant pool directly, but most of you will never use it; serp's high-level APIs completely abstract management of the constant pool. Any time a new constant is needed, serp will automatically add it to the pool while ensuring that no duplicates ever exist. Serp also does its best to manipulate the pool so that the effects of changing a constant are as expected: i.e. changing one instruction to use the string "bar" instead of "foo" will not affect other instructions that use the string "foo", but changing the name of a member field will instantly change all instructions that reference that field to use the new name.
• Instruction morphing. Dealing with the individual instructions that make up method code is the most difficult part of bytecode manipulation. To facilitate this process, most serp instruction representations have the ability to change their underlying low-level opcodes on the fly as the you modify the parameters of the instruction. For example, accessing the constant integer value 0 requires the opcode iconst0, while accessing the string constant "foo" requires a different opcode, ldc, followed by the constant pool index of "foo". In serp, however, there is only one instruction, constant. This instruction has setValue methods which use the given value to automatically determine the correct opcodes and arguments -- iconst0 for a value of 0 and ldc plus the proper constant pool index for the value of "foo".

Serp is not ideally suited to all applications. Here are a few disadvantages of serp:

• Speed. Serp is not built for speed. Though there are plans for performing incremental parsing, serp currently fully parses class files when a class is loaded, which is a slow process. Also, serp's insistence on full-time consistency between the low and high-level class structures slows down both access and mutator methods. These factors are less of a concern, though, when creating new classes at runtime (rather than modifying existing code), or when using serp as part of the compilation process. Serp excels in both of these scenarios.
• Memory. Serp's high-level structures for representing class bytecode are very memory-hungry.
• Multi-threaded modifications. The serp toolkit is not threadsafe. Multiple threads cannot safely make modifications to the same classes the same time.
• Project-level modifications. Changes made in one class in a serp project are not yet automatically propogated to other classes. However, there are plans to implement this, as well as plans to allow operations to modify bytecode based on specified patterns, similar to aspect-oriented programming.

The first step to using bytecode modification is to get familiar with the inner workings of class files. The class file format is covered here. Advanced manipulation involving changing or creating methods requires knowledge of the JVM instruction set, outlined here. Finally, learn to use the javap standard Java tool. Running javap -c on any class name will show you the opcodes of the class' methods -- this gives you an unlimited supply of examples to work from when writing your own opcodes.

The first class you should study in the serp bytecode toolkit is the serp.bytecode.Project type. From there, move onto the serp.bytecode.BCClass, and trace its APIs into serp.bytecode.BCField, serp.bytecode.BCMethod, and finally into actual serp.bytecode.Code.

Back to contents

Serp Version 1.15.0

New in 1.15.x:

• Ability to parse new Constant Pool Types (MethodHandle, MethodType, InvokeDynamic), new Instruction InvokeDynamic. These are necessary for classes leveraging Java 8 Lambda capabilities.
• Special case of blank method body.
• Performance improvement using StringBuilder instead of StringBuffer.
• Now compiled against Java 1.6 (was 1.3).

New in 1.14.x:

• Ability to easily reorder interface, field, and method declarations.
• APIs for finding declared methods based on covariant return types.

New in 1.13.x:

• Local variable scope now adapts to changes in the code block.

New in 1.12.x:

• Java 5 enhancements, most notably the ability to read and write class and member annotations.

New in 1.11.x:

• Serp now uses Apache Maven 2 for the build process.

New in 1.10.x:

• Exposed local variable type table to high-level bytecode APIs.
• Added low-level ConstantPoolTable for table-based fast access to the constant pool.

Download

• Serp distributions can be downloaded from http://serp.sourceforge.net/m2repo/net/sourceforge/serp/serp/.
• Anonymous CVS access (when prompted for a password, just hit the enter key):
  1. cvs -d:pserver:anonymous@serp.cvs.sourceforge.net:/cvsroot/serp login
  2. cvs -z3 -d:pserver:anonymous@serp.cvs.sourceforge.net:/cvsroot/serp co serp

Back to contents

License

Serp uses the BSD license. This is the most open license I could find; the goal is that regardless of your application -- commercial or non-commercial, open source or closed -- you can take advantage of the utilities serp provides, or even modify them to suit your needs... without any legal issues. Also, as in all open source, access to the source code means that you have the opportunity to fix blocking bugs without waiting for the next release. The following is the text of the license:

Copyright (c) 2002-2007, A. Abram White
All rights reserved.

Redistribution and use in source and binary forms, with or without 
modification, are permitted provided that the following conditions are met:

* Redistributions of source code must retain the above copyright notice, this 
  list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice, 
  this list of conditions and the following disclaimer in the documentation 
  and/or other materials provided with the distribution.
* Neither the name of 'serp' nor the names of its contributors may
  be used to endorse or promote products derived from this software without 
  specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED 
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE 
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES 
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON 
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
		

Back to contents