Parsing.py is used to generate LR(1) tables.
Those tables are output as Irken code, to be included with the parser.
The file tests/t20.scm implements the lexer generator and the LR(1) engine
  needed to do the actual parse.

still a work in progress, but here's the sequence so far:

1) write/edit lexer.py, generate "lexstep.scm"
2) write/edit grammar in "t0.g"
3) run "python gen_python_parser.py t0"
4) output will be t0.scm, to be included by tests/t20.scm