Two simple functions for rendering a Unicode string using ASCII characters only.

asciify.py

# /usr/bin/env python
# -*- encoding: utf-8 -*-
#
"""
Two simple functions for rendering a Unicode string using ASCII
characters only.

The only sensible applications are on words in a Latin-derived
alphabet (i.e., anything that could be rendered in a ISO-8859-*
character set); anything else will just be replaced by a string of
"unprintable" marks.
"""


def asciify_crudely(unistr, default='?'):
    # note that the entire docstring has to be *unicode*, otherwise
    # doctests fail inexplicably!
    u"""
    Render unicode string `unistr` using ASCII characters only.

    The rendering is quite crude in that any non-ASCII character is
    just replaced with the glyph `?`::

      >>> asciify_crudely(u'pâté')
      'p?t?'
      >>> asciify_crudely(u'PÂTÉ')
      'P?T?'

    The character to be substituted for untranslatable characters can
    be passed as second argument::

      >>> asciify_crudely(u'pâté', '*')
      'p*t*'
    """
    converted = []
    for unichr in iter(unistr):
        try:
            ch = unichr.decode('ascii')
        except:
            ch = default
        converted.append(chr(ord(ch)))
    return ''.join(converted)


def latinify(unistr, default='?'):
    # note that the entire docstring has to be *unicode*, otherwise
    # doctests fail inexplicably!
    u"""
    Render unicode string `unistr` using ASCII characters only.

    Latin letters with diacritical marks are substituted with their
    "bare" equivalent::

      >>> latinify(u'pâté')
      'pate'
      >>> latinify(u'PÂTÉ')
      'PATE'

    Letters which have no direct equivalent in the latin alphabet are
    replaced with the glyph `?`::

      >>> latinify(u'Sigurður Þórarinsson')
      'Sigur?ur ?orarinsson'

    The character to be substituted for untranslatable characters can
    be passed as second argument::

      >>> latinify(u'Sigurður Þórarinsson', '*')
      'Sigur*ur *orarinsson'
    """
    from unicodedata import name

    converted = []
    for unich in iter(unistr):
        try:
            ch = unich.decode('ascii')
        except:
            # deduce a latin letter equivalent from the Unicode data
            # point name; e.g., since `name(u'á') == 'LATIN SMALL
            # LETTER A WITH ACUTE'` translate `á` to `a`.  However, in
            # some cases the unicode name is still "LATIN LETTER"
            # although no direct equivalent in the Latin alphabeth
            # exists (e.g., Þ, "LATIN CAPITAL LETTER THORN") -- we can
            # avoid these cases by checking that the letter name is
            # composed of one letter only.
            what = name(unich).split()
            if what[0] == 'LATIN' and what[2] == 'LETTER' and len(what[3]) == 1:
                if what[1] == 'SMALL':
                    ch = what[3].lower()
                else: # what[1] == 'CAPITAL'
                    ch = what[3].upper()
            else:
                ch = default
        converted.append(chr(ord(ch)))
    return ''.join(converted)


if __name__ == '__main__':
    import doctest
    doctest.testmod(name='asciify',
                    optionflags=doctest.NORMALIZE_WHITESPACE)