1
|
|
|
<?php |
2
|
|
|
namespace vipnytt\RobotsTxtParser\Core\Directives; |
3
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Core\RobotsTxtInterface; |
5
|
|
|
use vipnytt\RobotsTxtParser\Core\Toolbox; |
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* Class UserAgent |
9
|
|
|
* |
10
|
|
|
* @package vipnytt\RobotsTxtParser\Core\Directives |
11
|
|
|
*/ |
12
|
|
|
class UserAgent implements DirectiveInterface, RobotsTxtInterface |
13
|
|
|
{ |
14
|
|
|
use Toolbox; |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* Sub directives white list |
18
|
|
|
*/ |
19
|
|
|
const SUB_DIRECTIVES = [ |
20
|
|
|
self::DIRECTIVE_ALLOW, |
21
|
|
|
self::DIRECTIVE_CACHE_DELAY, |
22
|
|
|
self::DIRECTIVE_COMMENT, |
23
|
|
|
self::DIRECTIVE_CRAWL_DELAY, |
24
|
|
|
self::DIRECTIVE_DISALLOW, |
25
|
|
|
self::DIRECTIVE_REQUEST_RATE, |
26
|
|
|
self::DIRECTIVE_ROBOT_VERSION, |
27
|
|
|
self::DIRECTIVE_VISIT_TIME, |
28
|
|
|
]; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* Directive |
32
|
|
|
*/ |
33
|
|
|
const DIRECTIVE = self::DIRECTIVE_USER_AGENT; |
34
|
|
|
|
35
|
|
|
/** |
36
|
|
|
* All User-agents declared |
37
|
|
|
* @var array |
38
|
|
|
*/ |
39
|
|
|
public $userAgents = []; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* Sub-directive Allow |
43
|
|
|
* @var DisAllow[] |
44
|
|
|
*/ |
45
|
|
|
public $allow = []; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* Sub-directive Cache-delay |
49
|
|
|
* @var CrawlDelay[] |
50
|
|
|
*/ |
51
|
|
|
public $cacheDelay = []; |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* Sub-directive Comment |
55
|
|
|
* @var Comment[] |
56
|
|
|
*/ |
57
|
|
|
public $comment = []; |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* Sub-directive Crawl-delay |
61
|
|
|
* @var CrawlDelay[] |
62
|
|
|
*/ |
63
|
|
|
public $crawlDelay = []; |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* Sub-directive Disallow |
67
|
|
|
* @var DisAllow[] |
68
|
|
|
*/ |
69
|
|
|
public $disallow = []; |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* Sub-directive Request-rate |
73
|
|
|
* @var RequestRate[] |
74
|
|
|
*/ |
75
|
|
|
public $requestRate = []; |
76
|
|
|
|
77
|
|
|
/** |
78
|
|
|
* Sub-directive Robot-version |
79
|
|
|
* @var RobotVersion[] |
80
|
|
|
*/ |
81
|
|
|
public $robotVersion = []; |
82
|
|
|
|
83
|
|
|
/** |
84
|
|
|
* Sub-directive Visit-time |
85
|
|
|
* @var VisitTime[] |
86
|
|
|
*/ |
87
|
|
|
public $visitTime = []; |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* Current User-agent(s) |
91
|
|
|
* @var array |
92
|
|
|
*/ |
93
|
|
|
protected $userAgent = []; |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* UserAgent constructor. |
97
|
|
|
*/ |
98
|
|
|
public function __construct() |
99
|
|
|
{ |
100
|
|
|
$this->set(); |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* Set new User-agent |
105
|
|
|
* |
106
|
|
|
* @param array $array |
107
|
|
|
* @return bool |
108
|
|
|
*/ |
109
|
|
|
public function set(array $array = [self::USER_AGENT]) |
110
|
|
|
{ |
111
|
|
|
$this->userAgent = array_map('mb_strtolower', $array); |
112
|
|
|
foreach ($this->userAgent as $userAgent) { |
113
|
|
|
if (!in_array($userAgent, $this->userAgents)) { |
114
|
|
|
$this->allow[$userAgent] = new DisAllow(self::DIRECTIVE_ALLOW); |
115
|
|
|
$this->cacheDelay[$userAgent] = new CrawlDelay(self::DIRECTIVE_CACHE_DELAY); |
116
|
|
|
$this->comment[$userAgent] = new Comment(); |
117
|
|
|
$this->crawlDelay[$userAgent] = new CrawlDelay(self::DIRECTIVE_CRAWL_DELAY); |
118
|
|
|
$this->disallow[$userAgent] = new DisAllow(self::DIRECTIVE_DISALLOW); |
119
|
|
|
$this->requestRate[$userAgent] = new RequestRate(); |
120
|
|
|
$this->robotVersion[$userAgent] = new RobotVersion(); |
121
|
|
|
$this->visitTime[$userAgent] = new VisitTime(); |
122
|
|
|
$this->userAgents[] = $userAgent; |
123
|
|
|
} |
124
|
|
|
} |
125
|
|
|
return true; |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
/** |
129
|
|
|
* Add |
130
|
|
|
* |
131
|
|
|
* @param string $line |
132
|
|
|
* @return bool |
133
|
|
|
*/ |
134
|
|
|
public function add($line) |
135
|
|
|
{ |
136
|
|
|
$result = []; |
137
|
|
|
$pair = $this->generateRulePair($line, self::SUB_DIRECTIVES); |
138
|
|
|
foreach ($this->userAgent as $userAgent) { |
139
|
|
|
switch ($pair['directive']) { |
140
|
|
|
case self::DIRECTIVE_ALLOW: |
141
|
|
|
$result[] = $this->allow[$userAgent]->add($pair['value']); |
142
|
|
|
break; |
143
|
|
|
case self::DIRECTIVE_CACHE_DELAY: |
144
|
|
|
$result[] = $this->cacheDelay[$userAgent]->add($pair['value']); |
145
|
|
|
break; |
146
|
|
|
case self::DIRECTIVE_COMMENT: |
147
|
|
|
$result[] = $this->comment[$userAgent]->add($pair['value']); |
148
|
|
|
break; |
149
|
|
|
case self::DIRECTIVE_CRAWL_DELAY: |
150
|
|
|
$result[] = $this->crawlDelay[$userAgent]->add($pair['value']); |
151
|
|
|
break; |
152
|
|
|
case self::DIRECTIVE_DISALLOW: |
153
|
|
|
$result[] = $this->disallow[$userAgent]->add($pair['value']); |
154
|
|
|
break; |
155
|
|
|
case self::DIRECTIVE_REQUEST_RATE: |
156
|
|
|
$result[] = $this->requestRate[$userAgent]->add($pair['value']); |
157
|
|
|
break; |
158
|
|
|
case self::DIRECTIVE_ROBOT_VERSION: |
159
|
|
|
$result[] = $this->robotVersion[$userAgent]->add($pair['value']); |
160
|
|
|
break; |
161
|
|
|
case self::DIRECTIVE_VISIT_TIME: |
162
|
|
|
$result[] = $this->visitTime[$userAgent]->add($pair['value']); |
163
|
|
|
break; |
164
|
|
|
} |
165
|
|
|
} |
166
|
|
|
return in_array(true, $result, true); |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
/** |
170
|
|
|
* Export rules |
171
|
|
|
* |
172
|
|
|
* @return array |
173
|
|
|
*/ |
174
|
|
|
public function export() |
175
|
|
|
{ |
176
|
|
|
$result = []; |
177
|
|
|
foreach ($this->userAgents as $userAgent) { |
178
|
|
|
$current = array_merge( |
179
|
|
|
$this->allow[$userAgent]->export(), |
180
|
|
|
$this->comment[$userAgent]->export(), |
181
|
|
|
$this->cacheDelay[$userAgent]->export(), |
182
|
|
|
$this->crawlDelay[$userAgent]->export(), |
183
|
|
|
$this->disallow[$userAgent]->export(), |
184
|
|
|
$this->requestRate[$userAgent]->export(), |
185
|
|
|
$this->robotVersion[$userAgent]->export(), |
186
|
|
|
$this->visitTime[$userAgent]->export() |
187
|
|
|
); |
188
|
|
|
if (!empty($current)) { |
189
|
|
|
$result[$userAgent] = $current; |
190
|
|
|
} |
191
|
|
|
} |
192
|
|
|
return empty($result) ? [] : [self::DIRECTIVE => $result]; |
193
|
|
|
} |
194
|
|
|
|
195
|
|
|
/** |
196
|
|
|
* Render |
197
|
|
|
* |
198
|
|
|
* @return string[] |
199
|
|
|
*/ |
200
|
|
|
public function render() |
201
|
|
|
{ |
202
|
|
|
$result = []; |
203
|
|
|
sort($this->userAgents); |
204
|
|
|
foreach ($this->userAgents as $userAgent) { |
205
|
|
|
$current = array_merge( |
206
|
|
|
$this->allow[$userAgent]->render(), |
207
|
|
|
$this->comment[$userAgent]->render(), |
208
|
|
|
$this->cacheDelay[$userAgent]->render(), |
209
|
|
|
$this->crawlDelay[$userAgent]->render(), |
210
|
|
|
$this->disallow[$userAgent]->render(), |
211
|
|
|
$this->requestRate[$userAgent]->render(), |
212
|
|
|
$this->robotVersion[$userAgent]->render(), |
213
|
|
|
$this->visitTime[$userAgent]->render() |
214
|
|
|
); |
215
|
|
|
if (!empty($current)) { |
216
|
|
|
$result = array_merge($result, [self::DIRECTIVE . ':' . $userAgent], $current); |
217
|
|
|
} |
218
|
|
|
} |
219
|
|
|
return $result; |
220
|
|
|
} |
221
|
|
|
} |
222
|
|
|
|