support/text_prettify.py

   1 import string
   2 from support.segment import segment
   3 from support.utilities import cat, lcat, sanitise
   4
   5
   6 def prettify(text, width=100):
   7     """Segment a text into words, then pack into lines, and combine the lines
   8     into a single string for printing."""
   9     return lcat(tpack(segment(text), width=width))
  10
  11
  12 def tpack(text, width=100):
  13     """Pack a list of words into lines, so long as each line (including
  14     intervening spaces) is no longer than _width_"""
  15     lines = [text[0]]
  16     for word in text[1:]:
  17         if len(lines[-1]) + 1 + len(word) <= width:
  18             lines[-1] += (' ' + word)
  19         else:
  20             lines += [word]
  21     return lines
  22
  23
  24 def depunctuate_character(c):
  25     """Record the punctuation of a character"""
  26     if c in string.ascii_uppercase:
  27         return 'UPPER'
  28     elif c in string.ascii_lowercase:
  29         return 'LOWER'
  30     else:
  31         return c
  32
  33
  34 def depunctuate(text):
  35     """Record the punctuation of a string, so it can be applied to a converted
  36     version of the string.
  37
  38     For example,
  39     punct = depunctuate(ciphertext)
  40     plaintext = decipher(sanitise(ciphertext))
  41     readable_plaintext = repunctuate(plaintext, punct)
  42     """
  43     return [depunctuate_character(c) for c in text]
  44
  45
  46 def repunctuate_character(letters, punctuation):
  47     """Apply the recorded punctuation to a character. The letters must be
  48     an iterator of base characters."""
  49     if punctuation == 'UPPER':
  50         return next(letters).upper()
  51     elif punctuation == 'LOWER':
  52         return next(letters).lower()
  53     else:
  54         return punctuation
  55
  56
  57 def repunctuate(text, punctuation):
  58     """Apply the recored punctuation to a sanitised string.
  59
  60     For example,
  61     punct = depunctuate(ciphertext)
  62     plaintext = decipher(sanitise(ciphertext))
  63     readable_plaintext = repunctuate(plaintext, punct)
  64     """
  65     letters = iter(sanitise(text))
  66     return cat(repunctuate_character(letters, p) for p in punctuation)