text_prettify.py

   1 from segment import segment
   2 from utilities import cat, sanitise
   3 import string
   4
   5
   6 def tpack(text, width=100):
   7     """Pack a list of words into lines, so long as each line (including
   8     intervening spaces) is no longer than _width_"""
   9     lines = [text[0]]
  10     for word in text[1:]:
  11         if len(lines[-1]) + 1 + len(word) <= width:
  12             lines[-1] += (' ' + word)
  13         else:
  14             lines += [word]
  15     return lines
  16
  17
  18 def depunctuate_character(c):
  19     """Record the punctuation of a character"""
  20     if c in string.ascii_uppercase:
  21         return 'UPPER'
  22     elif c in string.ascii_lowercase:
  23         return 'LOWER'
  24     else:
  25         return c
  26
  27
  28 def depunctuate(text):
  29     """Record the punctuation of a string, so it can be applied to a converted
  30     version of the string.
  31
  32     For example,
  33     punct = depunctuate(ciphertext)
  34     plaintext = decipher(sanitise(ciphertext))
  35     readable_plaintext = repunctuate(plaintext, punct)
  36     """
  37     return [depunctuate_character(c) for c in text]
  38
  39
  40 def repunctuate_character(letters, punctuation):
  41     """Apply the recorded punctuation to a character. The letters must be
  42     an iterator of base characters."""
  43     if punctuation == 'UPPER':
  44         return next(letters).upper()
  45     elif punctuation == 'LOWER':
  46         return next(letters).lower()
  47     else:
  48         return punctuation
  49
  50
  51 def repunctuate(text, punctuation):
  52     """Apply the recored punctuation to a sanitised string.
  53
  54     For example,
  55     punct = depunctuate(ciphertext)
  56     plaintext = decipher(sanitise(ciphertext))
  57     readable_plaintext = repunctuate(plaintext, punct)
  58     """
  59     letters = iter(sanitise(text))
  60     return cat(repunctuate_character(letters, p) for p in punctuation)