difflib ^[1] can do it.

Example from the docs:

>>> get_close_matches('appel', ['ape', 'apple', 'peach', 'puppy'])
['apple', 'ape']
>>> import keyword
>>> get_close_matches('wheel', keyword.kwlist)
['while']
>>> get_close_matches('apple', keyword.kwlist)
[]
>>> get_close_matches('accept', keyword.kwlist)
['except']

Check it out. It has other functions that can help you build something custom.

Levenshtein Python extension and C library.

https://github.com/ztane/python-Levenshtein/ ^[1]

The Levenshtein Python C extension module contains functions for fast computation of - Levenshtein (edit) distance, and edit operations - string similarity - approximate median strings, and generally string averaging - string sequence and set similarity It supports both normal and Unicode strings.

$ pip install python-levenshtein
...
$ python
>>> import Levenshtein
>>> help(Levenshtein.ratio)
ratio(...)
    Compute similarity of two strings.

    ratio(string1, string2)

    The similarity is a number between 0 and 1, it's usually equal or
    somewhat higher than difflib.SequenceMatcher.ratio(), becuase it's
    based on real minimal edit distance.

    Examples:
    >>> ratio('Hello world!', 'Holly grail!')
    0.58333333333333337
    >>> ratio('Brian', 'Jesus')
    0.0

>>> help(Levenshtein.distance)
distance(...)
    Compute absolute Levenshtein distance of two strings.

    distance(string1, string2)

    Examples (it's hard to spell Levenshtein correctly):
    >>> distance('Levenshtein', 'Lenvinsten')
    4
    >>> distance('Levenshtein', 'Levensthein')
    2
    >>> distance('Levenshtein', 'Levenshten')
    1
    >>> distance('Levenshtein', 'Levenshtein')
    0

As nosklo ^[1] said, use the difflib ^[2] module from the Python standard library.

The difflib module can return a measure of the sequences' similarity using the ratio() method of a SequenceMatcher() ^[3] object. The similarity is returned as a float in the range 0.0 to 1.0.

>>> import difflib

>>> difflib.SequenceMatcher(None, 'abcde', 'abcde').ratio()
1.0

>>> difflib.SequenceMatcher(None, 'abcde', 'zbcde').ratio()
0.80000000000000004

>>> difflib.SequenceMatcher(None, 'abcde', 'zyzzy').ratio()
0.0

Jellyfish ^[1] is a Python module which supports many string comparison metrics including phonetic matching. Pure Python implementations of Levenstein edit distance are quite slow compared to Jellyfish's implementation.

Example Usage:

import jellyfish

>>> jellyfish.levenshtein_distance('jellyfish', 'smellyfish')
2 
>>> jellyfish.jaro_distance('jellyfish', 'smellyfish')
0.89629629629629637
>>> jellyfish.damerau_levenshtein_distance('jellyfish', 'jellyfihs')
1
>>> jellyfish.metaphone('Jellyfish')
'JLFX'
>>> jellyfish.soundex('Jellyfish')
'J412'
>>> jellyfish.nysiis('Jellyfish')
'JALYF'
>>> jellyfish.match_rating_codex('Jellyfish')
'JLLFSH'`

I like nosklo's answer ^[1]; another method is the Damerau-Levenshtein distance ^[2]:

"In information theory and computer science, Damerau–Levenshtein distance is a 'distance' (string metric) between two strings, i.e., finite sequence of symbols, given by counting the minimum number of operations needed to transform one string into the other, where an operation is defined as an insertion, deletion, or substitution of a single character, or a transposition of two characters."

An implementation in Python from Wikibooks ^[3]:

def lev(a, b):
    if not a: return len(b)
    if not b: return len(a)
    return min(lev(a[1:], b[1:])+(a[0] != b[0]), \
    lev(a[1:], b)+1, lev(a, b[1:])+1)

More from Wikibooks, this gives you the length of the longest common substring (LCS) ^[4]:

def LCSubstr_len(S, T):
    m = len(S); n = len(T)
    L = [[0] * (n+1) for i in xrange(m+1)]
    lcs = 0
    for i in xrange(m):
        for j in xrange(n):
            if S[i] == T[j]:
                L[i+1][j+1] = L[i][j] + 1
                lcs = max(lcs, L[i+1][j+1])
    return lcs

There is also Google's own google-diff-match-patch ^[1] ("Currently available in Java, JavaScript, C++ and Python").

(Can't comment on it, since I have only used python's difflib myself)

Another alternative would be to use the recently released package FuzzyWuzzy ^[1]. The various functions supported by the package are also described in this blogpost ^[2].

I am using double-metaphone ^[1] which works like a charm.

An example:

>>> dm(u'aubrey')
('APR', '')
>>> dm(u'richard')
('RXRT', 'RKRT')
>>> dm(u'katherine') == dm(u'catherine')
True

Update: Jellyfish also has it. Comes under Phonetic encoding.

I've been using Fuzzy Wuzzy from Seat Geek with great success.

https://github.com/seatgeek/fuzzywuzzy

Specifically the token set ratio function...

They also did a great write up on the process of fuzzy string matching:

http://seatgeek.com/blog/dev/fuzzywuzzy-fuzzy-string-matching-in-python

Here's a python script for computing longest common substring in two words(may need tweaking to work for multi-word phrases):

def lcs(word1, word2):

    w1 = set(word1[i:j] for i in range(0, len(word1))
             for j in range(1, len(word1) + 1))

    w2 = set(word2[i:j] for i in range(0, len(word2))
             for j in range(1, len(word2) + 1))

    common_subs = w1.intersection(w2)

    sorted_cmn_subs = sorted([
        (len(str), str) for str in list(common_subs)
        ])

    return sorted_cmn_subs.pop()[1]

Heres the way how it can be done using Charicar's simhash, this is also suitable for long documents, it will detect 100% similarity also when you change order of words in documents too

http://blog.simpliplant.eu/calculating-similarity-between-text-strings-in-python/

Take a look at the Fuzzy ^[1] module. It has fast (written in C) based algorithms for soundex, NYSIIS and double-metaphone.

A good introduction can be found at: http://www.informit.com/articles/article.aspx?p=1848528