Download JAR files tagged by bytecode with all dependencies

All of our modules, somewhere along the way, use reflective structures wrapping java objects from various apis. Gwt uses JDT AST, Jre uses actual reflection, and our injection library uses a bytecode scanner to read in .class files. In order to abstract over the various apis we wrap, we have our own interface layer to describe java metadata, so we can create tools to utilize java structure independent of any given dependencies.

PowerMock allows you to unit test code normally regarded as untestable. For instance it is possible to mock static methods, remove static initializers, allow mocking without dependency injection and more. PowerMock works by bytecode manipulation. PowerMock also contain some utilities that gives you easier access to an objects internal state. PowerMock can be used to test otherwise untestable code and also to achieve a cleaner separation between test and production code.

This module contains a classpath scanner, and bytecode reader. It is useful in any java runtime environment, and allows java code to parse bytecode into a meta structure of classes, annotations, methods and fields. This meta structure is used throughout xapi to allow us to extend java structures in a cross-platform compliant way (exposing services and service objects so you can hack support for whatever runtime you need). It is especially useful for fast runtime injection, and as an extensible structure during codegen (which exposes mutable interfaces for our meta structure)

A bottom-up rewrite machine is a compiler construction tool that is often used in the compiler's back end to convert a tree-structured representation of a program into machine code -- or, in Java's case, bytecode. JBurg can also be used as a general-purpose dynamic programming engine. JBurg is descended from iburg-class BURGs, described in Fraser, Hanson, and Proebsting's paper, "Engineering a Simple, Efficient Code Generator Generator." JBurg brings similar O(N) minimum-cost tree rewriting capabilities to Java, and also allows the programmer to specify transitions between non-terminal states, that are significantly more powerful than iburg's transitive closures: JBurg transformation rules allow the transformation to inject additional program logic, which makes a JBurg specification more like a grammar than like a list of pattern-matching rules.

XApi is a dependency injection, service-oriented, polyglot java framework. It is comprised of standalone modules for use in web, desktop and mobile java applications. This codebase enables complete reuse of all code across your java stack, as well as the ability to easily define your own Platform, so you can build your own customized runtime environment to extend or replace existing platforms. It is currently geared towards servlets and gwt clients, but the target we are currently developing include: java, gwt [webkit, firefox, ie], javafx, playn [android, ios, flash], appengine & vert.x For Google Web Toolkit, we also include reflection support, more complete emulation of java.lang.Class, zero-overhead dependency injection, a range of code generation utilities, and even "magic-method injection" (for production compiles any method in the app can be swapped out with others based on configuration properties, or any AST mangling / code generating methods you wish to write). There is also a bytecode transformer in the works which will replace calls into the DI utility with direct references to static final factories, to allow the java compiler to completely erase all injection overhead in all java runtimes.