dict.py

   1 import re
   2 from nltk.wordnet import N,V,ADJ,ADV
   3
   4 from cache import cached
   5
   6 parts = [N, V, ADJ, ADV]
   7
   8 @cached
   9 def find_definition(word):
  10     # TODO: do fuzzy matching for misspelled words and its variations
  11     s = ""
  12     for part in parts:
  13         if word in part:
  14             s += "[%s]\n" % part.pos
  15             for sense in part[word]:
  16                 s += sense.gloss + "\n"
  17             s += "\n"
  18     return s
  19
  20 pat_quoted = re.compile('("[^"]*")')
  21 @cached
  22 def find_definition_html(word):
  23     """This is much like `find_definition_html' except the returned
  24     string is formatted as pretty HTML
  25     """
  26     s = ""
  27     for part in parts:
  28         if word in part:
  29             s += "<p><b>[%s]</b><ol>" % part.pos
  30             for sense in part[word]:
  31                 s += "<li>" + pat_quoted.sub(r'<i>\1</i>', sense.gloss)
  32                 s += "</li>"
  33             s += "</ol></p>"
  34     return s
  35
  36 if __name__ == '__main__':
  37     print "dog:\n", find_definition("dog")
  38     print "mitigate:\n", find_definition("mitigate")