randsent

#!/usr/bin/env python2

from __future__ import print_function

import argparse
from collections import defaultdict, deque
import random
import re
import sys

# Parse arguments.  You should not need to modify this.
parser = argparse.ArgumentParser()
parser.add_argument("grammar", help="path to grammar file")
parser.add_argument("count", type=int, help="number of sentences to generate", nargs='?', default=1)
parser.add_argument("-t", "--tree", action='store_true', help="pretty-print tree form instead of printing sentence")
parser.add_argument("--seed", type=int, default=0, help="RNG seed")
parser.add_argument("--min-size", type=int)
parser.add_argument("--max-size", type=int)
args = parser.parse_args()

# Create a random generator.
rng = random.Random(args.seed)

def parse_grammar(file_):
  rules = defaultdict(lambda: set([]))

  pattern = re.compile('^(?P<lhs>[^\s()#]+)\s+(?P<rhs>[^()#\n]*[^()#\n\s])')
  for line in file_:
    match = pattern.match(line)
    if match is None:
      continue
    groups = match.groupdict()
    lhs = groups['lhs']
    rhs = tuple(groups['rhs'].split())
    rules[lhs].add(rhs)
  assert 'START' in rules

  rules = { lhs: list(rhs) for (lhs, rhs) in rules.items() }

  return rules

def generate_sentence(grammar):
  queue = deque(['START'])

  while len(queue) > 0:
    sym = queue.popleft()

    # Is nonterminal?
    if sym in grammar:
      rhs = rng.choice(grammar[sym])
      queue.extendleft(reversed(rhs))
    else:
      yield sym

# Parse the grammar
with open(args.grammar, 'r') as grammar_file:
  grammar = parse_grammar(grammar_file)

for _ in xrange(args.count):
  while True:
    sent = list(generate_sentence(grammar))
    if (args.min_size is None or len(sent) >= args.min_size) and \
       (args.max_size is None or len(sent) <= args.max_size):
      break
  print(' '.join(sent))

# Local Variables:
# mode: python
# End: