|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Class SensitiveWord |
|
4
|
|
|
* @link https://www.icy2003.com/ |
|
5
|
|
|
* @author icy2003 <[email protected]> |
|
6
|
|
|
* @copyright Copyright (c) 2017, icy2003 |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace icy2003\php\ihelpers; |
|
9
|
|
|
|
|
10
|
|
|
use icy2003\php\icomponents\file\LocalFile; |
|
11
|
|
|
use icy2003\php\I; |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* 敏感词过滤 |
|
15
|
|
|
*/ |
|
16
|
|
|
class SensitiveWord |
|
17
|
|
|
{ |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* 敏感词树 |
|
21
|
|
|
* |
|
22
|
|
|
* @var array |
|
23
|
|
|
*/ |
|
24
|
|
|
protected $_map = []; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* 添加一个敏感词 |
|
28
|
|
|
* |
|
29
|
|
|
* @param string|array $content |
|
30
|
|
|
* |
|
31
|
|
|
* @return static |
|
32
|
|
|
*/ |
|
33
|
|
|
public function add($content) |
|
34
|
|
|
{ |
|
35
|
|
|
$words = Strings::split($content); |
|
|
|
|
|
|
36
|
|
|
$temp = &$this->_map; |
|
37
|
|
|
foreach ($words as $word) { |
|
38
|
|
|
if (false === isset($temp[$word])) { |
|
39
|
|
|
$temp[$word] = []; |
|
40
|
|
|
} |
|
41
|
|
|
$temp = &$temp[$word]; |
|
42
|
|
|
} |
|
43
|
|
|
$temp['isFind'] = true; |
|
44
|
|
|
return $this; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* 删除一个敏感词 |
|
49
|
|
|
* |
|
50
|
|
|
* @param string|array $content |
|
51
|
|
|
* |
|
52
|
|
|
* @return static |
|
53
|
|
|
*/ |
|
54
|
|
|
public function del($content) |
|
55
|
|
|
{ |
|
56
|
|
|
$words = is_string($content) ? Strings::split($content) : $content; |
|
57
|
|
|
$temp = &$this->_map; |
|
58
|
|
|
foreach ($words as $word) { |
|
59
|
|
|
if (isset($temp[$word])) { |
|
60
|
|
|
$temp = &$temp[$word]; |
|
61
|
|
|
} |
|
62
|
|
|
} |
|
63
|
|
|
$temp['isFind'] = false; |
|
64
|
|
|
return $this; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* 从头开始检测一个字符串是否是敏感词 |
|
69
|
|
|
* |
|
70
|
|
|
* - 发现敏感词,则返回该词 |
|
71
|
|
|
* - 没有敏感词,则返回 false |
|
72
|
|
|
* |
|
73
|
|
|
* @param string|array $content |
|
74
|
|
|
* @param boolean $findMore |
|
75
|
|
|
* |
|
76
|
|
|
* @return string|boolean |
|
77
|
|
|
*/ |
|
78
|
|
|
protected function _isStart($content, $findMore = true) |
|
79
|
|
|
{ |
|
80
|
|
|
$words = is_string($content) ? Strings::split($content) : $content; |
|
81
|
|
|
$temp = &$this->_map; |
|
82
|
|
|
$string = ''; |
|
83
|
|
|
$isFind = false; |
|
84
|
|
|
foreach ($words as $word) { |
|
85
|
|
|
if (isset($temp[$word])) { |
|
86
|
|
|
$string .= $word; |
|
87
|
|
|
if (true === I::get($temp[$word], 'isFind')) { |
|
88
|
|
|
$isFind = true; |
|
89
|
|
|
if (true === $findMore) { |
|
90
|
|
|
$temp = &$temp[$word]; |
|
91
|
|
|
continue; |
|
92
|
|
|
} else { |
|
93
|
|
|
return $string; |
|
94
|
|
|
} |
|
95
|
|
|
} else { |
|
96
|
|
|
$isFind = false; |
|
97
|
|
|
$temp = &$temp[$word]; |
|
98
|
|
|
continue; |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
|
|
return true === $isFind ? $string : false; |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
/** |
|
106
|
|
|
* 查找一段文本里所有敏感词 |
|
107
|
|
|
* |
|
108
|
|
|
* @param string|array $content |
|
109
|
|
|
* |
|
110
|
|
|
* @return array |
|
111
|
|
|
*/ |
|
112
|
|
|
public function find($content) |
|
113
|
|
|
{ |
|
114
|
|
|
$words = is_string($content) ? Strings::split($content) : $content; |
|
115
|
|
|
$strings = []; |
|
116
|
|
|
while (true) { |
|
117
|
|
|
if (empty($words)) { |
|
118
|
|
|
return $strings; |
|
119
|
|
|
} |
|
120
|
|
|
$string = $this->_isStart($words); |
|
121
|
|
|
if (is_string($string)) { |
|
122
|
|
|
$strings[] = $string; |
|
123
|
|
|
} |
|
124
|
|
|
array_shift($words); |
|
125
|
|
|
} |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* 返回替换敏感词后的文本 |
|
130
|
|
|
* |
|
131
|
|
|
* @param string|array $content |
|
132
|
|
|
* @param string $char |
|
133
|
|
|
* |
|
134
|
|
|
* @return string |
|
135
|
|
|
*/ |
|
136
|
|
|
public function replace($content, $char = '*') |
|
137
|
|
|
{ |
|
138
|
|
|
$words = $this->find($content); |
|
139
|
|
|
$content = implode('', $words); |
|
140
|
|
|
$array = []; |
|
141
|
|
|
foreach ($words as $word) { |
|
142
|
|
|
$array[$word] = Strings::repeat($char, Strings::length($word)); |
|
143
|
|
|
} |
|
144
|
|
|
return Strings::replace($content, $array); |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* 返回敏感词树 |
|
149
|
|
|
* |
|
150
|
|
|
* @return array |
|
151
|
|
|
*/ |
|
152
|
|
|
public function toArray() |
|
153
|
|
|
{ |
|
154
|
|
|
return $this->_map; |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
/** |
|
158
|
|
|
* 导出敏感词树到 php 文件 |
|
159
|
|
|
* |
|
160
|
|
|
* @param string|null $file |
|
161
|
|
|
* |
|
162
|
|
|
* @return void |
|
163
|
|
|
*/ |
|
164
|
|
|
public function toPhp($file = null) |
|
165
|
|
|
{ |
|
166
|
|
|
null === $file && $file = __DIR__ . '/' . date('YmdHis') . '.php'; |
|
167
|
|
|
$local = new LocalFile(); |
|
168
|
|
|
$array = Arrays::export($this->toArray(), true); |
|
169
|
|
|
$local->putFileContent($file, <<<EOT |
|
170
|
|
|
<?php |
|
171
|
|
|
|
|
172
|
|
|
return {$array}; |
|
173
|
|
|
|
|
174
|
|
|
EOT); |
|
175
|
|
|
} |
|
176
|
|
|
} |
|
177
|
|
|
|