Python implementation of Google Refine fingerprinting algorithms here: https://github.com/OpenRefine/OpenRefine/wiki/Clustering-In-Depth

fingerprint.py

# -*- coding: utf-8 -*-

import re, string
from unidecode import unidecode

PUNCTUATION = re.compile('[%s]' % re.escape(string.punctuation))

class Fingerprinter(object):
    '''
    Python implementation of Google Refine fingerprinting algorithm described here:
    https://github.com/OpenRefine/OpenRefine/wiki/Clustering-In-Depth

    Requires the unidecode module: https://github.com/iki/unidecode
    '''
    def __init__(self, string):
        self.string = self._preprocess(string)

    def _preprocess(self, string):
        '''
        Strip leading and trailing whitespace, lowercase the string, remove all punctuation,
        in that order.
        '''
        return PUNCTUATION.sub('', string.strip().lower())

    def _latinize(self, string):
        '''
        Replaces unicode characters with closest Latin equivalent. For example,
        Alejandro González Iñárritu becomes Alejando Gonzalez Inarritu.
        '''
        return unidecode(string.decode('utf-8'))

    def _unique_preserving_order(self, seq):
        '''
        Returns unique tokens in a list, preserving order. Fastest version found in this
        exercise: http://www.peterbe.com/plog/uniqifiers-benchmark
        '''
        seen = set()
        seen_add = seen.add
        return [x for x in seq if not (x in seen or seen_add(x))]
        
    def get_fingerprint(self):
        '''
        Gets conventional fingerpint.
        '''
        return self._latinize(' '.join(
            self._unique_preserving_order(
                sorted(self.string.split())
            )
        ))

    def get_ngram_fingerprint(self, n=1):
        '''
        Gets ngram fingerpint based on n-length shingles of the string.
        Default is 1.
        '''
        return self._latinize(''.join(
            self._unique_preserving_order(
                sorted([self.string[i:i + n] for i in range(len(self.string) - n + 1)])
            )
        ))

if __name__ == '__main__':
    f = Fingerprinter('Tom Cruise')
    print f.get_fingerprint()
    print f.get_ngram_fingerprint(n=1)

    f = Fingerprinter('Cruise, Tom')
    print f.get_fingerprint()
    print f.get_ngram_fingerprint(n=1)

    f = Fingerprinter('Paris')
    print f.get_fingerprint()
    print f.get_ngram_fingerprint(n=2)

For ngram fingerpint to match OpenRefine you need to remove all white spaces.
This is how I use it in pandas
df['SupplierName'].apply(lambda x: Fingerprinter(x).get_fingerprint())
df['SupplierName'].apply(lambda x: Fingerprinter(x.replace(" ", "")).get_ngram_fingerprint(n=2))

Thanks for sharing

Thanks for this. I created my own version based on this: https://gist.github.com/pietz/d6197f64c34d273a6d456d7b736c028d

Changes:

• Python 3.X compatible
• functional API
• bug fixed described by @safarend
• code-wise a little closer to OpenRefine docs
• added my own modified ngram version. It doesn't seem logical to me that the two Tom Cruise examples produce different results with ngram=2