|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* @package Application Utils |
|
4
|
|
|
* @subpackage ConvertHelper |
|
5
|
|
|
* @see \AppUtils\ConvertHelper\WordSplitter |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace AppUtils\ConvertHelper; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* @package Application Utils |
|
14
|
|
|
* @subpackage ConvertHelper |
|
15
|
|
|
* @author Sebastian Mordziol <[email protected]> |
|
16
|
|
|
*/ |
|
17
|
|
|
class WordSplitter |
|
18
|
|
|
{ |
|
19
|
|
|
private string $subject; |
|
20
|
|
|
private bool $removeDuplicates = false; |
|
21
|
|
|
private bool $sorting = false; |
|
22
|
|
|
private int $minWordLength = 0; |
|
23
|
|
|
private bool $duplicatesCaseInsensitive; |
|
24
|
|
|
|
|
25
|
|
|
public function __construct(string $subject) |
|
26
|
|
|
{ |
|
27
|
|
|
$this->subject = $subject; |
|
28
|
|
|
} |
|
29
|
|
|
|
|
30
|
|
|
public function setRemoveDuplicates(bool $remove=true, bool $caseInsensitive=false) : self |
|
31
|
|
|
{ |
|
32
|
|
|
$this->removeDuplicates = $remove; |
|
33
|
|
|
$this->duplicatesCaseInsensitive = $caseInsensitive; |
|
34
|
|
|
return $this; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
public function setSorting(bool $sorting=true) : self |
|
38
|
|
|
{ |
|
39
|
|
|
$this->sorting = $sorting; |
|
40
|
|
|
return $this; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
public function setMinWordLength(int $length) : self |
|
44
|
|
|
{ |
|
45
|
|
|
$this->minWordLength = $length; |
|
46
|
|
|
return $this; |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
public function split() : array |
|
50
|
|
|
{ |
|
51
|
|
|
$words = preg_split("/\W+/", $this->subject); |
|
52
|
|
|
|
|
53
|
|
|
$words = $this->filterEmpty($words); |
|
54
|
|
|
|
|
55
|
|
|
if($this->removeDuplicates) { |
|
56
|
|
|
$words = $this->filterDuplicates($words); |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
if($this->sorting) { |
|
60
|
|
|
usort($words, 'strnatcasecmp'); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
return $words; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
private function filterDuplicates(array $words) : array |
|
67
|
|
|
{ |
|
68
|
|
|
if($this->duplicatesCaseInsensitive) { |
|
69
|
|
|
return $this->filterDuplicatesCaseInsensitive($words); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
return array_unique($words); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
private function filterDuplicatesCaseInsensitive(array $array) : array |
|
76
|
|
|
{ |
|
77
|
|
|
return array_intersect_key( |
|
78
|
|
|
$array, |
|
79
|
|
|
array_unique( array_map( "strtolower", $array ) ) |
|
80
|
|
|
); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* @param string[] $words |
|
85
|
|
|
* @return string[] |
|
86
|
|
|
*/ |
|
87
|
|
|
private function filterEmpty(array $words) : array |
|
88
|
|
|
{ |
|
89
|
|
|
$keep = array(); |
|
90
|
|
|
|
|
91
|
|
|
foreach($words as $word) |
|
92
|
|
|
{ |
|
93
|
|
|
if(empty($word)) { |
|
94
|
|
|
continue; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
if(mb_strlen($word) < $this->minWordLength) { |
|
98
|
|
|
continue; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
$keep[] = $word; |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
return $keep; |
|
105
|
|
|
} |
|
106
|
|
|
} |
|
107
|
|
|
|
Mistralys /
application-utils
