src/mjacob/nltk/grammar/TagNonterminal.py

   1 # This Python file uses the following encoding: utf-8
   2 '''
   3 Created on May 18, 2011
   4
   5 @author: mjacob
   6 '''
   7 from nltk.grammar import Nonterminal
   8 import re
   9
  10 class InvalidTagNonterminalFormatException(Exception): pass
  11
  12 class TagNonterminal(Nonterminal):
  13     """class for expanding the metadata at a node in a tag _tree.
  14     allows easy querying and representation of label, foot status, and adjunction requirements"""
  15
  16     NO_ADJUNCTION = 'N'
  17     OBLIGATORY_ADJUNCTION = 'O'
  18     IS_FOOT = '*'
  19     NODE_TAG = re.compile('(\%s)?(\w+)(?:\.([%s%s]))?$' % (IS_FOOT, NO_ADJUNCTION, OBLIGATORY_ADJUNCTION))
  20
  21     def __init__(self, string):
  22         if not isinstance(string, str):
  23             raise InvalidTagNonterminalFormatException("unexpected type: %s %s" % (type(string), string))
  24         m = re.match(TagNonterminal.NODE_TAG, string)
  25         if not m:
  26             raise InvalidTagNonterminalFormatException("unrecognized tree node format: '%s'" % (string))
  27         self.__foot = m.group(1) == TagNonterminal.IS_FOOT
  28         Nonterminal.__init__(self, m.group(2))
  29         self.__rule = m.group(3) # N if NO ADJUNCTION, O of OBLIGATORY ADJUNCTION
  30
  31     def is_foot(self):
  32         return self.__foot
  33
  34     def no_adjunction(self):
  35         """true if adjunction is never allowed at the node"""
  36         return self.__rule == TagNonterminal.NO_ADJUNCTION
  37
  38     def obligatory_adjunction(self):
  39         """true if adjunction is MANDATORY at the node"""
  40         return self.__rule == TagNonterminal.OBLIGATORY_ADJUNCTION
  41
  42     def __str__(self):
  43         if self.__foot:
  44             f = TagNonterminal.IS_FOOT
  45         else:
  46             f = ""
  47         if self.no_adjunction():
  48             r = "." + TagNonterminal.NO_ADJUNCTION
  49         elif self.obligatory_adjunction():
  50             r = "." + TagNonterminal.OBLIGATORY_ADJUNCTION
  51         else:
  52             r = ""
  53         return "%s%s%s" % (f, self.symbol(), r)
  54
  55     def __repr__(self):
  56         return str(self)
  57
  58     def __eq__(self, other):
  59         return (type(self) is type(other)
  60             and Nonterminal.__eq__(self, other)
  61             and self.__foot == other.__foot
  62             and self.__rule == other.__rule)
  63
  64     def __hash__(self):
  65         return hash(self.__foot) ^ hash(self.__rule) ^ Nonterminal.__hash__(self)