|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* File containing the class {@see ConvertHelper_TextComparer}. |
|
4
|
|
|
* |
|
5
|
|
|
* @package Application Utils |
|
6
|
|
|
* @subpackage ConvertHelper |
|
7
|
|
|
* @see ConvertHelper_TextComparer |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
declare(strict_types=1); |
|
11
|
|
|
|
|
12
|
|
|
namespace AppUtils; |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Text comparison tool: can be used to calculate how |
|
16
|
|
|
* close two texts are from each other, using the |
|
17
|
|
|
* Levenshtein method. |
|
18
|
|
|
* |
|
19
|
|
|
* Converts the resulting match rating to a percentage |
|
20
|
|
|
* for easy processing. |
|
21
|
|
|
* |
|
22
|
|
|
* @package Application Utils |
|
23
|
|
|
* @subpackage ConvertHelper |
|
24
|
|
|
* @author Sebastian Mordziol <[email protected]> |
|
25
|
|
|
* |
|
26
|
|
|
* @see ConvertHelper::matchString() |
|
27
|
|
|
*/ |
|
28
|
|
|
class ConvertHelper_TextComparer implements Interface_Optionable |
|
29
|
|
|
{ |
|
30
|
|
|
use Traits_Optionable; |
|
31
|
|
|
|
|
32
|
|
|
const OPTION_MAX_LEVENSHTEIN_DISTANCE = 'maxLevenshtein'; |
|
33
|
|
|
const OPTION_PRECISION = 'precision'; |
|
34
|
|
|
|
|
35
|
|
|
public function getDefaultOptions() : array |
|
36
|
|
|
{ |
|
37
|
|
|
return array( |
|
38
|
|
|
self::OPTION_MAX_LEVENSHTEIN_DISTANCE => 10, |
|
39
|
|
|
self::OPTION_PRECISION => 1 |
|
40
|
|
|
); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
public function getMaxDistance() : int |
|
44
|
|
|
{ |
|
45
|
|
|
return $this->getIntOption(self::OPTION_MAX_LEVENSHTEIN_DISTANCE); |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
public function getPrecision() : int |
|
49
|
|
|
{ |
|
50
|
|
|
return $this->getIntOption(self::OPTION_PRECISION); |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* Sets the maximum Levensthein distance: results above this |
|
55
|
|
|
* value are ignored (will return a 0% match). |
|
56
|
|
|
* |
|
57
|
|
|
* @param int $distance |
|
58
|
|
|
* @return ConvertHelper_TextComparer |
|
59
|
|
|
*/ |
|
60
|
|
|
public function setMaxDistance(int $distance) : ConvertHelper_TextComparer |
|
61
|
|
|
{ |
|
62
|
|
|
return $this->setOption(self::OPTION_MAX_LEVENSHTEIN_DISTANCE, $distance); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* Sets the precision of the returned match percentage value. |
|
67
|
|
|
* |
|
68
|
|
|
* @param int $precision |
|
69
|
|
|
* @return ConvertHelper_TextComparer |
|
70
|
|
|
*/ |
|
71
|
|
|
public function setPrecision(int $precision) : ConvertHelper_TextComparer |
|
72
|
|
|
{ |
|
73
|
|
|
return $this->setOption(self::OPTION_PRECISION, $precision); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Calculates a percentage match of the source string with the target string. |
|
78
|
|
|
* |
|
79
|
|
|
* NOTE: The percentage is based on the maximum Levensthein distance |
|
80
|
|
|
* option. As such, the smaller the calculated distance, the higher |
|
81
|
|
|
* the percentage. The maximum distance equals to 0%. |
|
82
|
|
|
* |
|
83
|
|
|
* @param string $source |
|
84
|
|
|
* @param string $target |
|
85
|
|
|
* @return float |
|
86
|
|
|
*/ |
|
87
|
|
|
public function match(string $source, string $target) : float |
|
88
|
|
|
{ |
|
89
|
|
|
// avoid doing this via levenshtein |
|
90
|
|
|
if($source === $target) { |
|
91
|
|
|
return 100; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
$maxL = $this->getMaxDistance(); |
|
95
|
|
|
|
|
96
|
|
|
$diff = levenshtein($source, $target); |
|
97
|
|
|
if($diff > $maxL) { |
|
98
|
|
|
return 0; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
$percent = $diff * 100 / ($maxL + 1); |
|
102
|
|
|
return round(100 - $percent, $this->getPrecision()); |
|
103
|
|
|
} |
|
104
|
|
|
} |
|
105
|
|
|
|
Mistralys /
application-utils
