|
1
|
|
|
"""Class definitions for simple comparable types.""" |
|
2
|
|
|
|
|
3
|
|
|
import logging |
|
4
|
|
|
from difflib import SequenceMatcher |
|
5
|
|
|
|
|
6
|
|
|
from comparable.base import SimpleComparable |
|
7
|
|
|
|
|
8
|
|
|
|
|
9
|
|
|
class _Simple(SimpleComparable): # pylint: disable=W0223 |
|
10
|
|
|
|
|
11
|
|
|
"""SimpleComparable with common magic methods implemented.""" |
|
12
|
|
|
|
|
13
|
|
|
def __init__(self, value): |
|
14
|
|
|
self.value = value |
|
15
|
|
|
|
|
16
|
|
|
def __repr__(self): |
|
17
|
|
|
return self._repr(self.value) |
|
18
|
|
|
|
|
19
|
|
|
def __str__(self): |
|
20
|
|
|
return str(self.value) |
|
21
|
|
|
|
|
22
|
|
|
def __float__(self): |
|
23
|
|
|
return float(self.value) |
|
24
|
|
|
|
|
25
|
|
|
def __bool__(self): |
|
26
|
|
|
return bool(self.value) |
|
27
|
|
|
|
|
28
|
|
|
|
|
29
|
|
|
class Number(_Simple): |
|
30
|
|
|
|
|
31
|
|
|
"""Comparable positive number.""" |
|
32
|
|
|
|
|
33
|
|
|
threshold = 0.999 # 99.9% similar |
|
34
|
|
|
|
|
35
|
|
|
def __init__(self, value): |
|
36
|
|
|
super().__init__(value) |
|
37
|
|
|
if value < 0: |
|
38
|
|
|
raise ValueError("Number objects can only be positive") |
|
39
|
|
|
|
|
40
|
|
|
def equality(self, other): |
|
41
|
|
|
"""Get equality using floating point equality.""" |
|
42
|
|
|
return float(self) == float(other) |
|
43
|
|
|
|
|
44
|
|
|
def similarity(self, other): |
|
45
|
|
|
"""Get similarity as a ratio of the two numbers.""" |
|
46
|
|
|
numerator, denominator = sorted((self.value, other.value)) |
|
47
|
|
|
try: |
|
48
|
|
|
ratio = float(numerator) / denominator |
|
49
|
|
|
except ZeroDivisionError: |
|
50
|
|
|
ratio = 0.0 if numerator else 1.0 |
|
51
|
|
|
similarity = self.Similarity(ratio) |
|
52
|
|
|
return similarity |
|
53
|
|
|
|
|
54
|
|
|
|
|
55
|
|
|
class Text(_Simple): |
|
56
|
|
|
|
|
57
|
|
|
"""Comparable generic text.""" |
|
58
|
|
|
|
|
59
|
|
|
threshold = 0.83 # "Hello, world!" ~ "hello world" |
|
60
|
|
|
|
|
61
|
|
|
def equality(self, other): |
|
62
|
|
|
"""Get equality using string comparison.""" |
|
63
|
|
|
return str(self) == str(other) |
|
64
|
|
|
|
|
65
|
|
|
def similarity(self, other): |
|
66
|
|
|
"""Get similarity as a ratio of the two texts.""" |
|
67
|
|
|
ratio = SequenceMatcher(a=self.value, b=other.value).ratio() |
|
68
|
|
|
similarity = self.Similarity(ratio) |
|
69
|
|
|
return similarity |
|
70
|
|
|
|
|
71
|
|
|
|
|
72
|
|
|
class TextEnum(Text): |
|
73
|
|
|
|
|
74
|
|
|
"""Comparable case-insensitive textual enumeration.""" |
|
75
|
|
|
|
|
76
|
|
|
threshold = 1.0 # enumerations must match |
|
77
|
|
|
|
|
78
|
|
|
def similarity(self, other): |
|
79
|
|
|
"""Get similarity as a discrete ratio (1.0 or 0.0).""" |
|
80
|
|
|
ratio = 1.0 if (str(self).lower() == str(other).lower()) else 0.0 |
|
81
|
|
|
similarity = self.Similarity(ratio) |
|
82
|
|
|
return similarity |
|
83
|
|
|
|
|
84
|
|
|
|
|
85
|
|
|
class TextTitle(Text): |
|
86
|
|
|
|
|
87
|
|
|
"""Comparable case-insensitive textual titles.""" |
|
88
|
|
|
|
|
89
|
|
|
threshold = 0.93 # "The Cat and the Hat" ~ "cat an' the hat" |
|
90
|
|
|
|
|
91
|
|
|
ARTICLES = 'a', 'an', 'the' # stripped from the front |
|
92
|
|
|
JOINERS = '&', '+' # replaced with 'and' |
|
93
|
|
|
|
|
94
|
|
|
def __init__(self, value): |
|
95
|
|
|
super().__init__(value) |
|
96
|
|
|
self.stripped = self._strip(self.value) |
|
97
|
|
|
logging.debug("stripped %r to %r", self.value, self.stripped) |
|
98
|
|
|
|
|
99
|
|
|
@staticmethod |
|
100
|
|
|
def _strip(text): |
|
101
|
|
|
"""Strip articles/whitespace and remove case.""" |
|
102
|
|
|
text = text.strip() |
|
103
|
|
|
text = text.replace(' ', ' ') # remove duplicate spaces |
|
104
|
|
|
text = text.lower() |
|
105
|
|
|
for joiner in TextTitle.JOINERS: |
|
106
|
|
|
text = text.replace(joiner, 'and') |
|
107
|
|
|
for article in TextTitle.ARTICLES: |
|
108
|
|
|
if text.startswith(article + ' '): |
|
109
|
|
|
text = text[len(article) + 1:] |
|
110
|
|
|
break |
|
111
|
|
|
return text |
|
112
|
|
|
|
|
113
|
|
|
def similarity(self, other): |
|
114
|
|
|
"""Get similarity as a ratio of the stripped text.""" |
|
115
|
|
|
logging.debug("comparing %r and %r...", self.stripped, other.stripped) |
|
116
|
|
|
ratio = SequenceMatcher(a=self.stripped, b=other.stripped).ratio() |
|
117
|
|
|
similarity = self.Similarity(ratio) |
|
118
|
|
|
return similarity |
|
119
|
|
|
|
jacebrowning /
comparable
