yaccparse.py

#!/usr/bin/env python

import argparse
import collections
import re


args = argparse.ArgumentParser()
args.add_argument('input')

args = args.parse_args()


separators_reached = 0

symbol = re.compile('[_A-Za-z]+:')
first_production = re.compile('([^|][^{]*)({.*)?')


symbols = []
current_symbol = None
productions = []
in_code = False

for line in open(args.input):
    if line.strip() == '%%':
        separators_reached += 1
        continue

    if separators_reached != 1:
        continue

    line = line.strip()

    if symbol.match(line):
        if current_symbol:
            symbols.append((current_symbol, productions))
            productions = []

        current_symbol = line[:-1]
        continue

    if in_code:
        in_code = '}' not in line
        continue

    match = first_production.match(line)
    if match:
        (variables, code) = match.groups()

        variables = [ v for v in variables.split() if v != ';' ]
        if len(variables) > 0:
            assert len(productions) == 0
            productions = [ variables ]

        print('%s -> %s' % (current_symbol, productions))

        if code and '}' not in code:
            in_code = True

        continue


    print(line)

print(symbols)