Added a naive import cleaning module that looks for imported class name in code and if not found removes
the import declaration (It's not perfect in removing everything like Eclipse but it cannot harm as it may
only miss a redundant import but can't remove a useful one).
Also, changed java model to support multiple test files for a single java class, test files should be:
c_name.data (first test file)
c_name.1.data (second test file)
c_name.2.data (third test file)
and so on.
3 files changed
tree: 5ddbf3e911c2eb885ba17262d71d446c7ff3eb5c
  1. .abat-automerge
  2. .gitignore
  3. .hooks/
  4. AUTHORS
  5. INTERNALS
  6. LoxiGen.mk
  7. Makefile
  8. README.md
  9. TODO
  10. c_gen/
  11. generic_utils.py
  12. java_gen/
  13. lang_c.py
  14. lang_java.py
  15. lang_python.py
  16. loxi_front_end/
  17. loxi_ir.py
  18. loxi_utils/
  19. loxigen.py
  20. of_g.py
  21. openflow_input/
  22. py_gen/
  23. pyparsing.py
  24. tenjin.py
  25. test_data/
  26. utest/
README.md

Introduction

LoxiGen is a tool that generates OpenFlow protocol libraries for a number of languages. It is composed of a frontend that parses wire protocol descriptions and a backend for each supported language (currently C and Python, with Java on the way).

Usage

You can run LoxiGen directly from the repository. There's no need to install it, and it has no dependencies beyond Python 2.6+.

To generate the libraries for all languages:

make

To generate the library for a single language:

make c

The currently supported languages are c and python.

The generated libraries will be under the loxi_output directory. This can be changed with the LOXI_OUTPUT_DIR environment variable when using the Makefile.

Each generated library comes with its own set of documentation in the standard format for that language. Please see that documentation for more details on using the generated libraries.

Contributing

Please fork the repository on GitHub and send us a pull request. You might also be interested in the INTERNALS file which has notes about how LoxiGen works.