user-tools/levenshtein-distance

   1 #!/usr/bin/env python2.7
   2
   3 """
   4 =pod
   5
   6 =head1 NAME
   7
   8 levenshtein-distance - Calculate the Levenshtein distance of given strings
   9
  10 jaro-metric - Calculate the Jaro metric of given strings
  11
  12 jaro-winkler-metric - Calculate the Jaro-Winkler metric of given strings
  13
  14 =cut
  15
  16 """
  17
  18
  19 import Levenshtein
  20 import sys
  21 import os
  22
  23 try:
  24         str1 = sys.argv[1]
  25 except IndexError:
  26         raise Exception("""
  27 Usage:
  28   levenshtein-distance <string-1> [<string-2>]
  29   jaro-metric <string-1> [<string-2>]
  30   jaro-winkler-metric <string-1> [<string-2>]
  31 Environment:
  32   JARO_WINKLER_PREFIX_WEIGHT
  33 Reads strings to compare from stdin unless string-2 is given.
  34 Prints the resulting metric, or the <metric> TAB <string-2> if read from stdin.
  35 """)
  36
  37
  38 if len(sys.argv) > 2:
  39         str2 = sys.argv[2]
  40         read_stdin = False
  41 else:
  42         str2 = None
  43         read_stdin = True
  44
  45 jaro_winkler_prefix_weight = os.environ.get('JARO_WINKLER_PREFIX_WEIGHT')
  46 if jaro_winkler_prefix_weight is not None: jaro_winkler_prefix_weight = float(jaro_winkler_prefix_weight)
  47
  48
  49 if 'jaro' in sys.argv[0]:
  50         if 'winkler' in sys.argv[0]:
  51                 method = 'jaro_winkler'
  52         else:
  53                 method = 'jaro'
  54 else:
  55         method = 'distance'
  56
  57
  58 if read_stdin:
  59         while True:
  60                 str2 = sys.stdin.readline()
  61                 if str2 == '': break
  62                 str2 = str2.rstrip('\r\n')
  63
  64                 args = [str1, str2]
  65                 if jaro_winkler_prefix_weight is not None:
  66                         args.append(jaro_winkler_prefix_weight)
  67
  68                 print "%s\t%s" % (str(getattr(Levenshtein, method)(*args)), str2)
  69 else:
  70         args = [str1, str2]
  71         if jaro_winkler_prefix_weight is not None:
  72                 args.append(jaro_winkler_prefix_weight)
  73         print getattr(Levenshtein, method)(*args)