ProcessMovieName.py

import re
class ProcessMovieName:
  # 1 > we substring them, for the year. So incase if the string contains 
	#   any of the year mentioned, then it truncates it.
	listOfYears = range(1900,2020) # So i don't have to bother till 2020, instead 
	 							                 # we could just set the current year too.
	processedName ='';
	def cleanName(self,name):
		pattern = re.compile('[()\[\]]')
		name = pattern.sub('',name);
		pattern = re.compile("[']");
		name = pattern.sub('',name)
		name = re.sub('UNRATED','',name)
		name = re.sub('unrated','',name)

		
		pattern = re.compile("[._]")
		
		return pattern.sub(' ',name)



	

 	def processMovieName(self,name):
 		assert type(name) == str
  		[self.checkYear(name,years) for years in self.listOfYears]
  		return self.processedName;


 	def checkYear(self,name,year):

 		if name.find(str(year)) != -1:
 			name = name[0:name.find(str(year))]
 			self.processedName = self.cleanName(name);

  	def process(self,name):
  		
  		return self.processMovieName(name);