|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace detox\core; |
|
4
|
|
|
|
|
5
|
|
|
|
|
6
|
|
|
use detox\dataset\SetContract; |
|
7
|
|
|
use detox\source\Text; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* Class Words |
|
11
|
|
|
* @package detox\core |
|
12
|
|
|
* |
|
13
|
|
|
* @property Text text |
|
14
|
|
|
*/ |
|
15
|
|
|
class Words |
|
16
|
|
|
{ |
|
17
|
|
|
|
|
18
|
|
|
public const MAX_SCORE = 1; |
|
19
|
|
|
public const ASTERISKS_MIDDLE = 0.8; |
|
20
|
|
|
public const ASTERISKS_LEFT = 0.5; |
|
21
|
|
|
public const ASTERISKS_RIGHT = 0.6; |
|
22
|
|
|
|
|
23
|
|
|
protected $dataSet; |
|
24
|
|
|
protected $score = 0; |
|
25
|
|
|
protected $text; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Words constructor. |
|
29
|
|
|
* |
|
30
|
|
|
* @param SetContract $set |
|
31
|
|
|
* @param Text $text |
|
32
|
|
|
*/ |
|
33
|
|
|
public function __construct(SetContract $set, Text $text) |
|
34
|
|
|
{ |
|
35
|
|
|
$this->dataSet = $set; |
|
36
|
|
|
$this->text = $text; |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* Finds bad words from *Set and setting score/replace |
|
41
|
|
|
*/ |
|
42
|
|
|
public function processWords() : void |
|
43
|
|
|
{ |
|
44
|
|
|
// to match lower case letters in words set array |
|
45
|
|
|
$lowerSource = $this->addLowSpaces($this->text->getString()); |
|
46
|
|
|
/** |
|
47
|
|
|
* @var string $points |
|
48
|
|
|
* @var array $words |
|
49
|
|
|
*/ |
|
50
|
|
|
foreach ($this->dataSet->getWords() as $points => $words) { |
|
51
|
|
|
foreach ($words as $word) { |
|
52
|
|
|
if (mb_strpos($lowerSource, ' ' . $word . ' ') !== false) { |
|
53
|
|
|
$this->score += (float)$points; |
|
54
|
|
|
if ($this->text->isReplaceable()) { |
|
55
|
|
|
$this->replace($word); |
|
56
|
|
|
} |
|
57
|
|
|
} |
|
58
|
|
|
} |
|
59
|
|
|
} |
|
60
|
|
|
if ($this->score >= self::MAX_SCORE) { |
|
61
|
|
|
$this->score = self::MAX_SCORE; |
|
62
|
|
|
} |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* Finds bad words with asterisks and setting score/replace |
|
67
|
|
|
*/ |
|
68
|
|
|
public function processPatterns() : void |
|
69
|
|
|
{ |
|
70
|
|
|
$lowerSource = $this->addLowSpaces($this->text->getString()); |
|
71
|
|
|
if (preg_match_all('/\s(([\w]+)[\*]+([\w]+))\s/', $lowerSource, $matches) > 0) { |
|
72
|
|
|
$this->score += (self::ASTERISKS_MIDDLE * count($matches[0])); |
|
73
|
|
|
if ($this->text->isReplaceable()) { |
|
74
|
|
|
$this->replaceMatches($matches); |
|
75
|
|
|
} |
|
76
|
|
|
} |
|
77
|
|
|
$lowerSource = $this->addLowSpaces($this->text->getString()); |
|
78
|
|
|
if (preg_match_all('/\s(([\w]+)[\*]+)\s/', $lowerSource, $matches) > 0) { |
|
79
|
|
|
$this->score += (self::ASTERISKS_RIGHT * count($matches[0])); |
|
80
|
|
|
if ($this->text->isReplaceable()) { |
|
81
|
|
|
$this->replaceMatches($matches); |
|
82
|
|
|
} |
|
83
|
|
|
} |
|
84
|
|
|
$lowerSource = $this->addLowSpaces($this->text->getString()); |
|
85
|
|
|
if (preg_match_all('/\s([\*]+([\w]+))\s/', $lowerSource, $matches) > 0) { |
|
86
|
|
|
$this->score += (self::ASTERISKS_LEFT * count($matches[0])); |
|
87
|
|
|
if ($this->text->isReplaceable()) { |
|
88
|
|
|
$this->replaceMatches($matches); |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
/** |
|
94
|
|
|
* @param SetContract $set |
|
95
|
|
|
*/ |
|
96
|
|
|
public function setData(SetContract $set) : void |
|
97
|
|
|
{ |
|
98
|
|
|
$this->dataSet = $set; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
/** |
|
102
|
|
|
* @return float |
|
103
|
|
|
*/ |
|
104
|
|
|
public function getScore() : float |
|
105
|
|
|
{ |
|
106
|
|
|
return $this->score; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
/** |
|
110
|
|
|
* @param float $score |
|
111
|
|
|
*/ |
|
112
|
|
|
public function setScore(float $score) : void |
|
113
|
|
|
{ |
|
114
|
|
|
$this->score = $score; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* @param string $str |
|
119
|
|
|
* @return string |
|
120
|
|
|
*/ |
|
121
|
|
|
protected function addLowSpaces(string $str) : string |
|
122
|
|
|
{ |
|
123
|
|
|
return ' ' . mb_strtolower($str) . ' '; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
/** |
|
127
|
|
|
* @param string $phrase word or phrase to replace |
|
128
|
|
|
*/ |
|
129
|
|
|
protected function replace(string $phrase) : void |
|
130
|
|
|
{ |
|
131
|
|
|
// todo: slice the word with replacement to prevent WoRd -> word transformations |
|
132
|
|
|
$this->text->setString(str_replace($phrase, $this->text->getPrefix() . $this->text->getReplaceChars() . $this->text->getPostfix(), |
|
133
|
|
|
mb_strtolower($this->text->getString()))); |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
private function replaceMatches(array $matches) : void |
|
137
|
|
|
{ |
|
138
|
|
|
/** @var array $matches */ |
|
139
|
|
|
foreach ($matches[0] as $word) { |
|
140
|
|
|
$this->replace($word); |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
/** |
|
145
|
|
|
* Setter for convenient DI with Text object and it's properties |
|
146
|
|
|
* @param Text $text |
|
147
|
|
|
*/ |
|
148
|
|
|
public function setText(Text $text) : void |
|
149
|
|
|
{ |
|
150
|
|
|
$this->text = $text; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
/** |
|
154
|
|
|
* @return Text |
|
155
|
|
|
*/ |
|
156
|
|
|
public function getText() : Text |
|
157
|
|
|
{ |
|
158
|
|
|
return $this->text; |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
/** |
|
162
|
|
|
* @return SetContract |
|
163
|
|
|
*/ |
|
164
|
|
|
public function getDataSet() : SetContract |
|
165
|
|
|
{ |
|
166
|
|
|
return $this->dataSet; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
/** |
|
170
|
|
|
* @param SetContract $dataSet |
|
171
|
|
|
*/ |
|
172
|
|
|
public function setDataSet(SetContract $dataSet) : void |
|
173
|
|
|
{ |
|
174
|
|
|
$this->dataSet = $dataSet; |
|
175
|
|
|
} |
|
176
|
|
|
} |
arthurkushman /
detox
