# -*- coding: utf-8 -*-

  from BaseProcessor import baseProcessor
  import nltk

  import re

  from LiaTools import *
  class Orkis(baseProcessor):
      """ Processor for Orkis """
      def __init__(self,dirtyString):
          self.tagger=Tagger()
          self.phoner=Phoner()
          self.dico ={}
          self.string=dirtyString

  	print self.string

      def isReady(self):
          self.phoner.isReady()
          self.tagger.isReady()

      def __unicode__(self):
  	string = u""

          for word in self.dico:

              string += ( unicode(word.decode("utf-8")))+unicode (u";")

              for lemWord in self.dico[word][0]:

                  string += (unicode(lemWord.decode("utf-8"))+ unicode(u" "))

              string +=u";"

              for phonWord in self.dico[word][1]:

                  string += (unicode(phonWord.decode("utf-8"))+ unicode(u" "))

              string+=u"
  "

          return string

      def clean(self):
          stopword=StopWord()
          self.string=stopword.RemoveStopList(self.string)
      def insertLem(self):
          self.cleanString=self.tagger.clean(self.string)
          taggedString=self.tagger.tagg(self.cleanString)
          self.tableLem = taggedString.rstrip().split("
  ")
          for line in taggedString.rstrip().split("
  "):

  	    if not re.match(r's>',line):
              	table = line.rstrip().split(" ")
              	if not table[0] in self.dico :
                  	self.dico[table[0]]=[set(),set()]
              	self.dico[table[0]][0].add(table[2])

      def insertPhon(self):
          phonedString=self.phoner.phon(self.cleanString)
          self.tablephon= phonedString.rstrip().split("
  ")
          for line in phonedString.rstrip().split("
  "):

  	    if not re.match(r's>',line):
                  table = line.rstrip().split(" ")
                  if table[0] in self.dico:
                      self.dico[table[0]][1].add(table[1])

      def getDico(self):
          self.clean()
          self.insertLem()
          self.insertPhon()

          table=[]

          for i,v in self.dico.iteritems():    

              if not re.match(r"<s>",i):

                  list=[]          
                  list.append(i)

                  for indice in v[0]:

                      list.append(indice) 

                  for indice in v[1]:  

                      list.append(indice)
                  ligne= " ".join(list)     

                  table.append(ligne) 
          return "
  ".join(table)