grammar.py

""" MODULE TO PARSE PCFG GRAMMAR """

from collections import defaultdict
from math import fsum

class PCFG(object):
	"""pcfg representation"""

	def __init__(self, grammarfile):
		self.startsym = None
		self.lhsrules = defaultdict(list)
		self.rhsrules = defaultdict(list)
		self.readrules(grammarfile)


	def readrules(self, gramfile):

		for line in gramfile:	
			line = line.strip()

			# if not commented and line exists,
			if not line.startswith('#') and line:
			
				if "->" in line:
					# if line is a rule 
					rule = self.parse(line)
					lhs, rhs, prob = rule
					# add rule to dict for rhs and lhs
					self.lhsrules[lhs].append(rule)
					self.rhsrules[rhs].append(rule)

				else: 
					# line is a start symbol
					start, prob = line.rsplit(';')
					self.startsym = start.strip()

	def parse(self, rule):

		lhs, rest = rule.split('->')
		lhs = lhs.strip()

		rhs, prob = rest.rsplit(";")
		rhstup = tuple(rhs.strip().split())
		prob = float(prob)

		# print('lhs, rhs, prob \n', lhs, rhstup, prob)
		# dont forget return stmt
		return lhs, rhstup, prob

	def verifyGram(self):
		"""
		Return True if the grammar is a valid PCFG in CNF.
		Otherwise return False. 
		"""

		# for each in self.lhsrules:
		# 	print("(lhs)->", each)

		# for each in self.rhsrules:
		# 	print("(rhs)->", each)

		for key,values in self.lhsrules.items():
			lhs = values[0][0]
			#print(lhs)
			rhs=[]
			for i in range(len(values)):
			    rhs.append(values[i][2])

			#print(round(sum(rhs)))
			test = round(fsum(rhs))
			if test==1:
			    return True
			else:
			    return False



if __name__ == '__main__':

	# with open('probcfg.pcfg', 'r') as gramfile:
	with open('pcfg.cfg', 'r') as gramfile:
		grammar = PCFG(gramfile)

	if grammar.verifyGram():
		print("Yes, grammar is a valid PCFG in CNF")
	else:
		print("No, grammar is not a valid PCFG in CNF")