Absimpa is a Java class library that provides a recursive decent parser. It is an experiment in abstraction and Java generics. It tries to include only the essentials of what makes out a parser and to leave out, or rather leave free for specification, what is not relevant for the parsing process.
Approdictio provides implementations of dictionaries that allow approximate lookup. When looking up a word, all words are returned that are approximately equal to the word given. The definition of approximately equal depends on the metric provided to define the distance between two words. An implementation of the Levensthein Metric with customizable edit costs is part of the package.
Clig uses a simple description file to create C-code to interprete the typical *NIX command line as well as an up-to-date usage-message and a manual page skeleton. It supports Flag, String, Int, Long, Float and Double types, with ranges, defaults, and more. The generated C-Code is ANSI but has been reported to work with C++. It is self contained code which does not depend on any library other than libc. Included is a TCL-only TCL-package to instrument your TCL scripts the same way as your C-programs.
monq.jfa is a class library for fast and flexible text filtering with regular expressions. In contrast to java.util.regex, monq.jfa allows a regular expression to be bound to an action that is automatically called whenever a match is spotted in an input stream. In addition, it can combine several ten thousand regex/action pairs into one machinery (called a "deterministic finite automaton" or DFA) for fast parallel text filtering. A general purpose TCP server and associated classes in monq.net allow the bigger DFAs to be run as servers.
Xmldego provides XML parsing and/or transformation by binding callbacks to XML parsing events in the context of an XML element. While SAX, for example, calls your callbacks on every start tag, Xmldego will do so only for the child axis path you request. Xmldego is particularly useful when only smaller bits of a much larger XML file need to be processed, because only interest in the relevant bits need to be coded. In contrast to a DOM parser, much larger XML files can be processed, because the whole file is not loaded into memory.
Yruba provides a rule system similar to make or ant for the shell (bash). It provides a clear separation between a list of dependencies that must be up-to-date before the current task can be performed, an explicit test that checks whether the target is really out-of-date, and a command that finally makes the target. Everything is plain bash syntax, so there is no new command language to learn.
Is there a chance to get "use" links into the graph. In particular uses of other classes as parameters of public methods would be very interesting.