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
|
|
|
* 1st level checking discrete words with spaces |
48
|
|
|
* @var string $points |
49
|
|
|
* @var array $words |
50
|
|
|
*/ |
51
|
|
|
foreach ($this->dataSet->getWords() as $points => $words) { |
52
|
|
|
foreach ($words as $word) { |
53
|
|
|
if (mb_strpos($lowerSource, ' ' . $word . ' ') !== false) { |
54
|
|
|
$this->score += (float)$points; |
55
|
|
|
if ($this->text->isReplaceable()) { |
56
|
|
|
$this->replace($word); |
57
|
|
|
} |
58
|
|
|
} |
59
|
|
|
} |
60
|
|
|
} |
61
|
|
|
if ($this->score < self::ASTERISKS_MIDDLE) { |
62
|
|
|
// 2nd level checking the worst (3 highest levels) words with probability to be mutated ex.: shittyass, crappish, unfuckkish etc |
63
|
|
|
$slice = array_slice($this->dataSet->getWords(), 0, 3); |
64
|
|
|
foreach ($slice as $points => $words) { |
65
|
|
|
foreach ($words as $word) { |
66
|
|
|
if (mb_strpos($lowerSource, $word) !== false) { |
67
|
|
|
$this->score += (float)$points; |
68
|
|
|
if ($this->text->isReplaceable()) { |
69
|
|
|
$this->replace($word); |
70
|
|
|
} |
71
|
|
|
} |
72
|
|
|
} |
73
|
|
|
} |
74
|
|
|
} |
75
|
|
|
if ($this->score >= self::MAX_SCORE) { |
76
|
|
|
$this->score = self::MAX_SCORE; |
77
|
|
|
} |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
/** |
81
|
|
|
* Finds bad words with asterisks and setting score/replace |
82
|
|
|
*/ |
83
|
|
|
public function processPatterns() : void |
84
|
|
|
{ |
85
|
|
|
$lowerSource = $this->addLowSpaces($this->text->getString()); |
86
|
|
|
if (preg_match_all('/\s(([\w]+)[\*]+([\w]+))\s/', $lowerSource, $matches) > 0) { |
87
|
|
|
$this->score += (self::ASTERISKS_MIDDLE * count($matches[0])); |
88
|
|
|
if ($this->text->isReplaceable()) { |
89
|
|
|
$this->replaceMatches($matches); |
90
|
|
|
} |
91
|
|
|
} |
92
|
|
|
$lowerSource = $this->addLowSpaces($this->text->getString()); |
93
|
|
|
if (preg_match_all('/\s(([\w]+)[\*]+)\s/', $lowerSource, $matches) > 0) { |
94
|
|
|
$this->score += (self::ASTERISKS_RIGHT * count($matches[0])); |
95
|
|
|
if ($this->text->isReplaceable()) { |
96
|
|
|
$this->replaceMatches($matches); |
97
|
|
|
} |
98
|
|
|
} |
99
|
|
|
$lowerSource = $this->addLowSpaces($this->text->getString()); |
100
|
|
|
if (preg_match_all('/\s([\*]+([\w]+))\s/', $lowerSource, $matches) > 0) { |
101
|
|
|
$this->score += (self::ASTERISKS_LEFT * count($matches[0])); |
102
|
|
|
if ($this->text->isReplaceable()) { |
103
|
|
|
$this->replaceMatches($matches); |
104
|
|
|
} |
105
|
|
|
} |
106
|
|
|
} |
107
|
|
|
|
108
|
|
|
/** |
109
|
|
|
* @param SetContract $set |
110
|
|
|
*/ |
111
|
|
|
public function setData(SetContract $set) : void |
112
|
|
|
{ |
113
|
|
|
$this->dataSet = $set; |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
/** |
117
|
|
|
* @return float |
118
|
|
|
*/ |
119
|
|
|
public function getScore() : float |
120
|
|
|
{ |
121
|
|
|
return $this->score; |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* @param float $score |
126
|
|
|
*/ |
127
|
|
|
public function setScore(float $score) : void |
128
|
|
|
{ |
129
|
|
|
$this->score = $score; |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
/** |
133
|
|
|
* @param string $str |
134
|
|
|
* @return string |
135
|
|
|
*/ |
136
|
|
|
protected function addLowSpaces(string $str) : string |
137
|
|
|
{ |
138
|
|
|
return ' ' . mb_strtolower($str) . ' '; |
139
|
|
|
} |
140
|
|
|
|
141
|
|
|
/** |
142
|
|
|
* @param string $phrase word or phrase to replace |
143
|
|
|
*/ |
144
|
|
|
protected function replace(string $phrase) : void |
145
|
|
|
{ |
146
|
|
|
// todo: slice the word with replacement to prevent WoRd -> word transformations |
147
|
|
|
$this->text->setString(str_replace($phrase, $this->text->getPrefix() . $this->text->getReplaceChars() . $this->text->getPostfix(), |
148
|
|
|
mb_strtolower($this->text->getString()))); |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
private function replaceMatches(array $matches) : void |
152
|
|
|
{ |
153
|
|
|
/** @var array $matches */ |
154
|
|
|
foreach ($matches[0] as $word) { |
155
|
|
|
$this->replace($word); |
156
|
|
|
} |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
/** |
160
|
|
|
* Setter for convenient DI with Text object and it's properties |
161
|
|
|
* @param Text $text |
162
|
|
|
*/ |
163
|
|
|
public function setText(Text $text) : void |
164
|
|
|
{ |
165
|
|
|
$this->text = $text; |
166
|
|
|
} |
167
|
|
|
|
168
|
|
|
/** |
169
|
|
|
* @return Text |
170
|
|
|
*/ |
171
|
|
|
public function getText() : Text |
172
|
|
|
{ |
173
|
|
|
return $this->text; |
174
|
|
|
} |
175
|
|
|
|
176
|
|
|
/** |
177
|
|
|
* @return SetContract |
178
|
|
|
*/ |
179
|
|
|
public function getDataSet() : SetContract |
180
|
|
|
{ |
181
|
|
|
return $this->dataSet; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* @param SetContract $dataSet |
186
|
|
|
*/ |
187
|
|
|
public function setDataSet(SetContract $dataSet) : void |
188
|
|
|
{ |
189
|
|
|
$this->dataSet = $dataSet; |
190
|
|
|
} |
191
|
|
|
} |