Because unicode is fun, there are multiple characters for spaces, dashes, etc. Some of which have HTML equivalents, such as "NO-BREAK SPACE" which I don't want to respect, as the data source may supply that, but the output destination doesn't expect that.

fix_unicode_fun_characters.py

from __future__ import unicode_literals

import unicodedata
import ftfy
import re

"""
Taken from:
https://www.cs.tut.fi/~jkorpela/chars/spaces.html
https://en.wikipedia.org/wiki/Whitespace_character#Unicode
http://www.unicode.org/reports/tr44/#GC_Values_Table
http://www.fileformat.info/info/unicode/category/Zs/list.htm
http://www.fileformat.info/info/unicode/category/Zl/list.htm
http://www.fileformat.info/info/unicode/category/Zp/list.htm
http://www.fileformat.info/info/unicode/category/Cf/list.htm
http://www.fileformat.info/info/unicode/category/So/list.htm
http://www.fileformat.info/info/unicode/category/Pd/list.htm
http://www.charbase.com/
"""

collapses = (
    unicodedata.lookup('MONGOLIAN VOWEL SEPARATOR'),
    unicodedata.lookup('ZERO WIDTH SPACE'),
    unicodedata.lookup('ZERO WIDTH NO-BREAK SPACE'),
    unicodedata.lookup('NARROW NO-BREAK SPACE'),
    unicodedata.lookup('ZERO WIDTH NON-JOINER'),
    unicodedata.lookup('ZERO WIDTH JOINER'),
    unicodedata.lookup('WORD JOINER'),
)
collapses_chars = "".join(collapses)
collapses_piped = "|".join(collapses)
collapses_re = re.compile(r'%s' % collapses_piped, re.UNICODE)

spaces = (
    unicodedata.lookup('NO-BREAK SPACE'),
    unicodedata.lookup('NARROW NO-BREAK SPACE'),
    unicodedata.lookup('EN QUAD'),
    unicodedata.lookup('EM QUAD'),
    unicodedata.lookup('EN SPACE'),
    unicodedata.lookup('EM SPACE'),
    unicodedata.lookup('THREE-PER-EM SPACE'),
    unicodedata.lookup('FOUR-PER-EM SPACE'),
    unicodedata.lookup('SIX-PER-EM SPACE'),
    unicodedata.lookup('FIGURE SPACE'),
    unicodedata.lookup('PUNCTUATION SPACE'),
    unicodedata.lookup('THIN SPACE'),
    unicodedata.lookup('HAIR SPACE'),
    unicodedata.lookup('MEDIUM MATHEMATICAL SPACE'),
    unicodedata.lookup('IDEOGRAPHIC SPACE'),
    unicodedata.lookup('IDEOGRAPHIC HALF FILL SPACE'),
    unicodedata.lookup('SYMBOL FOR SPACE'),
    unicodedata.lookup('TAG SPACE'),
)
spaces_chars = "".join(spaces)
spaces_piped = "|".join(spaces)
spaces_re = re.compile(r'%s' % spaces_piped, re.UNICODE)

dashes = (
    unicodedata.lookup('OGHAM SPACE MARK'),
    unicodedata.lookup('HYPHEN'),
    unicodedata.lookup('SOFT HYPHEN'),
    unicodedata.lookup('HYPHEN-MINUS'),
    unicodedata.lookup('NON-BREAKING HYPHEN'),
    unicodedata.lookup('FIGURE DASH'),
    unicodedata.lookup('EN DASH'),
    unicodedata.lookup('EM DASH'),
    unicodedata.lookup('HORIZONTAL BAR'),
    unicodedata.lookup('MINUS SIGN'),
    unicodedata.lookup('SMALL EM DASH'),
    unicodedata.lookup('SMALL HYPHEN-MINUS'),
    unicodedata.lookup('FULLWIDTH HYPHEN-MINUS'),
    unicodedata.lookup('TAG HYPHEN-MINUS'),
)
dashes_chars = "".join(dashes)
dashes_piped = "|".join(dashes)
dashes_re = re.compile(r'%s' % dashes_piped, re.UNICODE)

seps = (
    unicodedata.lookup('LINE SEPARATOR'),
    unicodedata.lookup('PARAGRAPH SEPARATOR'),
)
seps_chars = "".join(seps)
seps_piped = "|".join(seps)
seps_re = re.compile(r'%s' % seps_piped, re.UNICODE)


def ftfy_further(value):
    workers = (
        (collapses_re, ""),
        (spaces_re, " "),
        (dashes_re, "-"),
        (seps_re, "\r\n"),
    )
    for iteration, handler in enumerate(workers, start=1):
        regex, replacement = handler
        value = regex.sub(replacement, value)
    return value


bad_text = """
this is some {collapses} collapsed characters.
this is some {spaces} fun space characters.
this is some {dashes} dashes characters.
this is some {seps} break characters.
""".format(collapses=collapses_chars,
           spaces=spaces_chars,
           dashes=dashes_chars,
           seps=seps_chars)
ftfy_text = ftfy.fix_text(bad_text)
result = ftfy_further(bad_text)
print(bad_text)
print(ftfy_text)
print(result)