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