1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* |
4
|
|
|
* @Name : similar-text |
5
|
|
|
* @Programmer : Akpé Aurelle Emmanuel Moïse Zinsou |
6
|
|
|
* @Date : 2019-04-01 |
7
|
|
|
* @Released under : https://github.com/manuwhat/similar-text/blob/master/LICENSE |
8
|
|
|
* @Repository : https://github.com/manuwhat/similar |
9
|
|
|
* |
10
|
|
|
**/ |
11
|
|
|
|
12
|
|
|
|
13
|
|
|
namespace EZAMA{ |
14
|
|
|
|
15
|
|
|
class complexCommonTextSimilarities extends complexCommonTextSimilaritiesHelper |
16
|
|
|
{ |
17
|
|
|
public static function strippedUrl($a, $b) |
18
|
|
|
{ |
19
|
|
|
if (self::isUrl($a, $domain) && is_string($b)) { |
20
|
|
|
return $domain === trim($b); |
21
|
|
|
} elseif (self::isUrl($b, $domain) && is_string($a)) { |
22
|
|
|
return $domain === trim($a); |
23
|
|
|
} else { |
24
|
|
|
return false; |
25
|
|
|
} |
26
|
|
|
} |
27
|
|
|
public static function areStems($a, $b) |
28
|
|
|
{ |
29
|
|
|
if (!is_string($a) || !is_string($b)) { |
30
|
|
|
return false; |
31
|
|
|
} |
32
|
|
|
|
33
|
|
|
$a = self::getParts(self::strtolower($a)); |
34
|
|
|
$b = self::getParts(self::strtolower($b)); |
35
|
|
|
foreach ((array) $a as $index=>$word) { |
36
|
|
|
if (!self::haveSameRoot($word, $b[$index])) { |
37
|
|
|
return false; |
38
|
|
|
} |
39
|
|
|
} |
40
|
|
|
return true; |
41
|
|
|
} |
42
|
|
|
|
43
|
|
|
public static function wordsAddedOrRemoved($a, $b) |
44
|
|
|
{ |
45
|
|
|
if (!is_string($a) || !is_string($b)) { |
46
|
|
|
return false; |
47
|
|
|
} |
48
|
|
|
$filter = function($v) { |
49
|
|
|
return !(ctype_space($v)); |
50
|
|
|
}; |
51
|
|
|
self::filter($a, $b, $filter, true); |
52
|
|
|
return self::waorDiff($a, $b, count($a), count($b)); |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
|
56
|
|
|
public static function punctuationChangesOccured($a, $b, $insensitive = true, $considerSpace = true) |
57
|
|
|
{ |
58
|
|
|
$filter = function($v) use ($considerSpace) { |
59
|
|
|
return $considerSpace ? !(ctype_space($v) || ctype_punct($v)) : !ctype_punct($v); |
60
|
|
|
}; |
61
|
|
|
if (!is_string($a) || !is_string($b)) { |
62
|
|
|
return false; |
63
|
|
|
} |
64
|
|
|
self::filter($a, $b, $filter, $insensitive); |
65
|
|
|
return empty(array_diff($a, $b)); |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
|
69
|
|
|
public static function acronymOrExpanded($a, $b) |
70
|
|
|
{ |
71
|
|
|
if (!is_string($a) || !is_string($b)) { |
72
|
|
|
return false; |
73
|
|
|
} |
74
|
|
|
$filter = function($v) { |
75
|
|
|
return !(ctype_space($v[0]) || ctype_punct($v[0])); |
76
|
|
|
}; |
77
|
|
|
|
78
|
|
|
self::filter($a, $b, $filter, true, true); |
79
|
|
|
return self::aoeStemming($a, $b); |
80
|
|
|
} |
81
|
|
|
} |
82
|
|
|
} |
83
|
|
|
|