scripts/anonymize

   1 #!/usr/bin/env python3
   2
   3 import sys
   4 import argparse
   5 import os
   6 import json
   7 import re
   8
   9 parser = argparse.ArgumentParser()
  10 parser.add_argument("words_file", help="File that contains the words to (de)anonymize")
  11 parser.add_argument("--ignore-missing", "-i", action="store_true", help="treat missing file as empty list")
  12 parser.add_argument("--deanonymize", "-d", action="store_true", help="deanonymize")
  13 config = parser.parse_args()
  14
  15 alphabet="abcdefghijklmnopqrstuvwxyz"
  16
  17 try:
  18     key = os.environ["ANONYMIZE_KEY"].lower()
  19     assert all([k in alphabet for k in key])
  20 except KeyError:
  21     print("Please set ANONYMIZE_KEY as environment variable with only letters", file=sys.stderr)
  22     sys.exit(1)
  23
  24 if not os.path.isfile(config.words_file):
  25     if config.ignore_missing:
  26         print(sys.stdin.read(), end="")
  27         sys.exit(0)
  28     else:
  29         print("Could not find words file", file=sys.stderr)
  30         sys.exit(1)
  31
  32 words = json.load(open(config.words_file))
  33
  34 if any([len(word) > len(key) for word in words]):
  35     print("The key needs to be at least as long as the longest word in the list (append to existing one to keep already mangled words)", file=sys.stderr)
  36     sys.exit(1)
  37
  38 order = -1 if config.deanonymize else 1
  39
  40 def replace(match):
  41     name = match.group()
  42     result = []
  43     for k in range(len(name)):
  44         if name[k].lower() not in alphabet:
  45             result.append(name[k])
  46         else:
  47             key_index = alphabet.index(key[k])
  48             letter_index = alphabet.index(name[k].lower())
  49             new_letter = alphabet[(letter_index + order * key_index) % len(alphabet)]
  50             if name[k].lower() != name[k]:
  51                 new_letter = new_letter.upper()
  52             result.append(new_letter)
  53     return ''.join(result)
  54
  55 regexp = re.compile("(" + '|'.join([r'(\b' + w + r'\b)' for w in words]) + ")")
  56 print(regexp.sub(replace, sys.stdin.read()), end="")