Source Code for Module parser_course.small_parsers.small_grammar

 1  try: 
 2      set    
 3  except: 
 4      from sets import Set # Not Python 2.4 
 5      set = Set 
 6       
 7  import multivalued_dictionary, supertuple 
 8   
 9 -class Grammar(object): 
10 -        def __init__(self,start_cat="s"): 
11               self.start_cat = start_cat 
12               self.compute_nonterminals() 
13               self.compute_preterminals() 
14   
15 -        def display (self): 
16              print 'Start Cat: %s' % self.start_cat 
17              print 'Rules: %s' % self.rules 
18              print 'Lexicon: %s' % self.lexicon 
19   
20 -        def compute_nonterminals (self): 
21              self.nonterminals = self.rules.keys() 
22   
23 -        def compute_preterminals (self): 
24              flat =[] 
25              for L in self.lexicon.values(): 
26                  flat += L 
27              self.preterminals = set(flat) 
28   
29 -        def get_parts_of_speech(self, word): 
30              """word is a string.  Return a list of 
31                 parts of speech for this word""" 
32              try: 
33                  return self.lexicon[word] 
34              except KeyError: 
35                  raise KeyError, "Unknown word: '%s' " % (word,) 
36   
37   
38   
39 -class toy_grammar (Grammar): 
40 -        def __init__(self): 
41              self.rules={"s": [["np","vp"],["aux","np","vp"]], 
42                          "np": [["det","nom"],["nom"],["np","pp"]], 
43                          "vp":[["v"],["v","np"]], 
44                          "pp":[["p","np"]], 
45                          "nom":[["n"]]} 
46              self.lexicon={"walk":["v","n"], 
47                            "the":["det"], 
48                            "dogs":["n"], 
49                            "dog":["n"], 
50                            "park":["v","n"], 
51                            "has":["aux"], 
52                            "have":["aux"], 
53                            "in":["p"]} 
54              Grammar.__init__(self, "s") 
55   
56   
57 -class not_so_toy_grammar (Grammar): 
58 -        def __init__(self): 
59              self.start_cat="s" 
60              self.rules={"s": [["np","vp"]], 
61                          "vp":[["vbz","np"],["vbz","np","pp"],["vbz","np","pp","pp"]], 
62                          "vpg":[["vbg","np"]], 
63                          "np": [["dt","nbar"],["nbar"],["np","cc","np"]], 
64                          "nbar":[["ap","nbar"],["nbar","pp"],["vpg"],["n"],["n","n"]], 
65                          "pp":[["p","np"]], 
66                          "ap":[["rb","a"],["a"]], 
67  } 
68              self.lexicon={"sees":["vbz"], 
69                            "the":["dt"], 
70                            "a":["dt"], 
71                            "agency":["n"], 
72                            "use":["n"], 
73                            "codes":["n"], 
74                            "way":["n"], 
75                            "mail":["n"], 
76                            "volume":["n"], 
77                            "labor":["n"], 
78                            "costs":["n"], 
79                            "of":["p"], 
80                            "as":["p"], 
81                            "widespread":["a"], 
82                            "growing":["a"], 
83                            "handling":["vbg"], 
84                            "controlling":["vbg"], 
85                            "rapidly":["rb"], 
86                            "and":["cc"]} 
87              Grammar.__init__(self) 
88