Source Code for Module nltk.wordnet.browse

  1  # Natural Language Toolkit: Wordnet Interface: Wordnet Text Mode Browser 
  2  # 
  3  # Copyright (C) 2001-2008 NLTK Project 
  4  # Author: Steven Bird <[email protected]> 
  5  #         Jussi Salmela <[email protected]> (modifications) 
  6  # URL: <http://nltk.org> 
  7  # For license information, see LICENSE.TXT 
  8   
  9  """Natural Language Toolkit: Wordnet Interface: Wordnet Text Mode Browser 
 10  See also the NLTK Wordnet Graphical Browser in nltk_contrib.wordnet 
 11  """ 
 12   
 13  from textwrap import TextWrapper 
 14  from random import randint 
 15   
 16  from util import * 
 17  from dictionary import * 
 18   
 19  tw = TextWrapper(subsequent_indent="    ") 
 20   
 21 -def show(synsets, index): 
 22      return "%d %s;" % (index, synsets[index][0]) 
 23   
 24 -def print_gloss(synsets, index): 
 25      print index, "\n".join(tw.wrap(synsets[index].gloss)) 
 26   
 27 -def print_all_glosses(synsets): 
 28      for index in range(len(synsets)): 
 29          print_gloss(synsets, index) 
 30   
 31 -def print_all(synsets): 
 32      for index in range(len(synsets)): 
 33          print show(synsets, index), 
 34      print 
 35   
 36 -def print_help(): 
 37      print "="*60 
 38      print "Lookup a word by typing it and finishing with Enter." 
 39      print "Reserved words -- letters and numbers used as browser commands --" 
 40      print "can be searched by preceeding them with an asterisk *." 
 41      print 
 42      print "Words have numbered senses, pick a sense by typing a number." 
 43      print 
 44      print "Commands are a letter followed by Enter:" 
 45      print "    d=down, u=up, g=gloss, s=synonyms, a=all-senses" 
 46      print "    v=verbose, r=random, q=quit" 
 47      print 
 48      print "Choose POS with: N=nouns, V=verbs, J=adjectives, R=adverbs" 
 49      print "="*60 
 50   
 51 -def new_word(word): 
 52      D = None 
 53      for pos,sec in ((N,"N"), (V,"V"), (ADJ,"J"), (ADV,"R")): 
 54          if word in pos: 
 55              if not D: D = pos 
 56              print sec, 
 57              print_all(pos[word]) 
 58      if D: synsets = D[word] 
 59      else: 
 60          print "Word '%s' not found! Choosing a random word." % word 
 61          D = N 
 62          synsets = random_synset(D) 
 63          print "N", 
 64          print_all(N[synsets[0][0]]) 
 65      return D, synsets 
 66   
 67 -def random_synset(D): 
 68      return D[randint(0,len(D)-1)] 
 69   
 70   
 71 -def browse(word=" ", index=0): 
 72      """ 
 73      Browse WordNet interactively, starting from the specified word, and 
 74      navigating the WordNet hierarchy to synonyms, hypernyms, hyponyms, and so on. 
 75   
 76      @type word: C{string} 
 77      @param word: the word to look up in WordNet 
 78      @type index: C{int} 
 79      @param index: the sense number of this word to use (optional) 
 80      """ 
 81      print "Wordnet browser (type 'h' for help)" 
 82      D, synsets = new_word(word) 
 83   
 84      while True: 
 85          if index >= len(synsets): 
 86              index = 0 
 87          input = '' 
 88          while input == '': 
 89              if synsets: 
 90                  prompt = "%s_%d/%d>" % (synsets[index][0], index, len(synsets)) 
 91                  input = raw_input(prompt) 
 92              else: 
 93                  input = raw_input("> ")  # safety net 
 94   
 95          # word lookup 
 96          if len(input) > 1 and not input.isdigit(): 
 97              if input[0] == "*": 
 98                  word = input[1:] 
 99              else: 
100                  word = input.lower() 
101              D, synsets = new_word(word) 
102              index = 0 
103   
104          # sense selection 
105          elif input.isdigit(): 
106              if int(input) < len(synsets): 
107                  index = int(input) 
108                  print_gloss(synsets, index) 
109              else: 
110                  print "There are %d synsets" % len(synsets) 
111   
112          # more info 
113          elif input == "a": 
114              print_all(synsets) 
115          elif input == "g": 
116              print_gloss(synsets, index) 
117          elif input == "v": 
118              print_all_glosses(synsets) 
119          elif input == "s": 
120              print "Synonyms:", ' '.join(word for word in synsets[index]) 
121   
122          # choose part-of-speech 
123          elif input in "NVJR": 
124              ind = "NVJR".index(input) 
125              pos = [N, V, ADJ, ADV][ind] 
126              s = ["noun", "verb", "adjective", "adverb"][ind] 
127              if word in pos: 
128                  D = pos 
129                  synsets = D[word] 
130              else: 
131                  print "No " + s + " sense found" 
132   
133          # navigation 
134          elif input == "r": 
135              synsets = random_synset(D) 
136          elif input == "u": 
137              try: 
138                  hypernyms = synsets[index][HYPERNYM] 
139                  hypernyms[0] 
140                  synsets = hypernyms 
141                  print_all(synsets) 
142                  index = 0 
143              except IndexError: 
144                  print "Cannot go up" 
145          elif input == "d": 
146              try: 
147                  hyponyms = synsets[index][HYPONYM] 
148                  hyponyms[0] 
149                  synsets = hyponyms 
150                  print_all(synsets) 
151                  index = 0 
152              except IndexError: 
153                  print "Cannot go down" 
154   
155          # miscellany 
156          elif input == "h" or input == "?": 
157              print_help() 
158          elif input == "q": 
159              print "Goodbye" 
160              break 
161   
162          else: 
163              print "Unrecognised command: %s" % input 
164              print "Type 'h' for help" 
165   
166 -def demo(): 
167      print_help() 
168      print 
169      browse() 
170   
171  if __name__ == '__main__': 
172      demo() 
173   
174  __all__ = ["demo"] 
175