This project is mostly abandoned! If you actually want to get this Software to run on your machine or build something on top of it please get in touch: stefan_troost {at} web {dot} de. I hate it when I find an abandoned piece of software that I would really like to use but can't figure it out how to get the darn thing to run. This way, we can figure it out together if you want.
This is the github page for my Bachelor's thesis, "An RPython JIT for Parsing Expression Grammars".
The current title for the JIT, "pypeg" might change soon, since "pypeg" is already the name of an interpreter based on parsing expression grammars.
A more detailed overview of the Project, it's application and results is available under https://github.com/sktroost/PyPeg/blob/master/pypeg/pypeg_thesis.pdf
However, here is a basic rundown:
--What is this--
A Parsing Expression Grammar is a type of analytic formal grammar, introduced by Bryan Ford in 2004. Applications for PEGs include pattern-matching, making them an alternative to regular expressions used in tools like grep. Due to the expressive power of PEGs they can be also used to parse context-free languages.
In Practice, this means we can use PEGs to search for a pattern in a string of text (i.e. search a text for valid URLs), or parse a string of a certain format (i.e. JSON, C).
--How does it work--
LPeg is another interpreter that implements PEGs by parsing them into bytecode and executing that through a virtual machine (VM). PyPeg contains a VM that uses the same bytecode generated by LPeg, but executing them significantly faster at sufficiently large filesizes (see https://github.com/sktroost/PyPeg/tree/master/pypeg/plots/blogpostplots )
This speedup is achieved by using PyPy's RPython toolchain, which allows our python program to be compiled into C-Code, and our Interpreter to generate x86-Assembly code for better performance.
--How do I install it?--
Currently, this software requires a lot of dependencies and is not very user friendly. Current requirements include
-lua5.2 (installed in ubuntu by opening a terminal and typing sudo apt-get install lua5.2) -libffi-dev (pypy requirement) (installed in ubuntu by opening a terminal and typing sudo apt-get install libffi-dev) -pypy source (https://bitbucket.org/pypy/pypy/src)
A more detailed installation guide may be available soon.
--On what platforms does it run?-- Currently x86-Linux only.