/
__init__.py
120 lines (100 loc) · 3.78 KB
/
__init__.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
import re
import math
import operator
import collections
import Levenshtein
__version__ = (0, 0, 5)
_non_word_re = re.compile(r'[^\w, ]+')
__all__ = ('FuzzySet',)
class FuzzySet(object):
" Fuzzily match a string "
def __init__(self, iterable=(), gram_size_lower=2, gram_size_upper=3, use_levenshtein=True):
self.exact_set = set()
self.match_dict = collections.defaultdict(list)
self.items = {}
self.use_levenshtein = use_levenshtein
self.gram_size_lower = gram_size_lower
self.gram_size_upper = gram_size_upper
for i in range(gram_size_lower, gram_size_upper + 1):
self.items[i] = []
for value in iterable:
self.add(value)
def add(self, value):
value = value.lower()
if value in self.exact_set:
return False
for i in range(self.gram_size_lower, self.gram_size_upper + 1):
self.__add(value, i)
def __add(self, value, gram_size):
items = self.items[gram_size]
idx = len(items)
items.append(0)
grams = _gram_counter(value, gram_size)
norm = math.sqrt(sum(x**2 for x in grams.values()))
for gram, occ in grams.items():
self.match_dict[gram].append((idx, occ))
items[idx] = (norm, value)
self.exact_set.add(value)
def __getitem__(self, value):
value = value.lower()
if value in self.exact_set:
return [(1, value)]
for i in range(self.gram_size_upper, self.gram_size_lower - 1, -1):
results = self.__get(value, i)
if results is not None:
return results
raise KeyError(value)
def __get(self, value, gram_size):
matches = collections.defaultdict(float)
grams = _gram_counter(value, gram_size)
items = self.items[gram_size]
norm = math.sqrt(sum(x**2 for x in grams.values()))
for gram, occ in grams.items():
for idx, other_occ in self.match_dict.get(gram, ()):
matches[idx] += occ * other_occ
if not matches:
return None
# cosine similarity
results = [(match_score / (norm * items[idx][0]), items[idx][1])
for idx, match_score in matches.items()]
results.sort(reverse=True, key=operator.itemgetter(0))
if self.use_levenshtein:
results = [(_distance(matched, value), matched)
for _, matched in results[:50]]
results.sort(reverse=True, key=operator.itemgetter(0))
return [result for result in results
if result[0] == results[0][0]]
def get(self, key, default=None):
try:
return self[key]
except KeyError:
return default
def _distance(str1, str2):
distance = Levenshtein.distance(str1, str2)
if len(str1) > len(str2):
return 1 - float(distance) / len(str1)
else:
return 1 - float(distance) / len(str2)
def _gram_counter(value, gram_size=2):
result = collections.defaultdict(int)
for value in _iterate_grams(value, gram_size):
result[value] += 1
return result
def _iterate_grams(value, gram_size=2):
simplified = '-' + _non_word_re.sub('', value.lower()) + '-'
len_diff = gram_size - len(simplified)
if len_diff > 0:
value += '-' * len_diff
for i in range(len(simplified) - gram_size + 1):
yield simplified[i:i + gram_size]
def _other_test():
with open('./origin_cities') as cities:
for line in cities:
result = f.get(line.strip())
if result is None:
print "{}: Could not find".format(line.strip())
elif isinstance(result, list):
print "{}: {}".format(line.strip(), result)
if __name__ == '__main__':
pass
#_other_test()