1
|
|
|
<?php |
2
|
|
|
namespace vipnytt\RobotsTxtParser; |
3
|
|
|
|
4
|
|
|
use Composer\CaBundle\CaBundle; |
5
|
|
|
|
6
|
|
|
/** |
7
|
|
|
* Class UriClient |
8
|
|
|
* |
9
|
|
|
* @package vipnytt\RobotsTxtParser |
10
|
|
|
*/ |
11
|
|
|
class UriClient extends TxtClient |
12
|
|
|
{ |
13
|
|
|
/** |
14
|
|
|
* User-agent |
15
|
|
|
*/ |
16
|
|
|
const CURL_USER_AGENT = 'RobotsTxtParser-VIPnytt/2.0 (+https://github.com/VIPnytt/RobotsTxtParser/blob/master/README.md)'; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* Base uri |
20
|
|
|
* @var string |
21
|
|
|
*/ |
22
|
|
|
private $base; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* Header parser |
26
|
|
|
* @var Parser\HeaderParser |
27
|
|
|
*/ |
28
|
|
|
private $headerParser; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* RequestClient timestamp |
32
|
|
|
* @var int |
33
|
|
|
*/ |
34
|
|
|
private $time; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Status code |
38
|
|
|
* @var int|null |
39
|
|
|
*/ |
40
|
|
|
private $statusCode; |
|
|
|
|
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* Effective uri |
44
|
|
|
* @var string |
45
|
|
|
*/ |
46
|
|
|
private $effectiveUri; |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* Cache-Control max-age |
50
|
|
|
* @var int |
51
|
|
|
*/ |
52
|
|
|
private $maxAge; |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* Robots.txt contents |
56
|
|
|
* @var string |
57
|
|
|
*/ |
58
|
|
|
private $contents; |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* Robots.txt character encoding |
62
|
|
|
* @var string |
63
|
|
|
*/ |
64
|
|
|
private $encoding; |
|
|
|
|
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* RequestClient constructor. |
68
|
|
|
* |
69
|
|
|
* @param string $baseUri |
70
|
|
|
* @param array $curlOptions |
71
|
|
|
* @param int|null $byteLimit |
72
|
|
|
*/ |
73
|
|
|
public function __construct($baseUri, array $curlOptions = [], $byteLimit = self::BYTE_LIMIT) |
74
|
|
|
{ |
75
|
|
|
$this->base = $this->urlBase($this->urlEncode($baseUri)); |
76
|
|
|
if ($this->request($curlOptions) === false) { |
77
|
|
|
$this->time = time(); |
78
|
|
|
$this->effectiveUri = $this->base . self::PATH; |
79
|
|
|
$this->statusCode = null; |
80
|
|
|
$this->contents = ''; |
81
|
|
|
$this->encoding = self::ENCODING; |
82
|
|
|
$this->maxAge = 0; |
83
|
|
|
} |
84
|
|
|
parent::__construct($this->base, $this->statusCode, $this->contents, $this->encoding, $this->effectiveUri, $byteLimit); |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
/** |
88
|
|
|
* cURL request |
89
|
|
|
* |
90
|
|
|
* @param array $options |
91
|
|
|
* @return bool |
92
|
|
|
*/ |
93
|
|
|
private function request($options = []) |
94
|
|
|
{ |
95
|
|
|
$this->headerParser = new Parser\HeaderParser(); |
96
|
|
|
$ch = curl_init(); |
97
|
|
|
curl_setopt_array($ch, [ |
98
|
|
|
CURLOPT_AUTOREFERER => true, |
99
|
|
|
CURLOPT_CAINFO => CaBundle::getSystemCaRootBundlePath(), |
100
|
|
|
CURLOPT_CONNECTTIMEOUT => 30, |
101
|
|
|
CURLOPT_ENCODING => 'identity', |
102
|
|
|
CURLOPT_FAILONERROR => false, |
103
|
|
|
CURLOPT_FOLLOWLOCATION => true, |
104
|
|
|
CURLOPT_FTPSSLAUTH => CURLFTPAUTH_DEFAULT, |
105
|
|
|
CURLOPT_HEADER => false, |
106
|
|
|
CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_NONE, |
107
|
|
|
CURLOPT_IPRESOLVE => CURL_IPRESOLVE_WHATEVER, |
108
|
|
|
CURLOPT_NOBODY => false, |
109
|
|
|
CURLOPT_MAXREDIRS => self::MAX_REDIRECTS, |
110
|
|
|
CURLOPT_RETURNTRANSFER => true, |
111
|
|
|
CURLOPT_SSL_VERIFYHOST => 2, |
112
|
|
|
CURLOPT_SSL_VERIFYPEER => true, |
113
|
|
|
//CURLOPT_SSL_VERIFYSTATUS => true, // PHP 7.0.7 |
114
|
|
|
CURLOPT_TIMEOUT => 120, |
115
|
|
|
CURLOPT_USERAGENT => self::CURL_USER_AGENT, |
116
|
|
|
CURLOPT_USERPWD => 'anonymous:', |
117
|
|
|
]); |
118
|
|
|
curl_setopt_array($ch, $options); |
119
|
|
|
curl_setopt($ch, CURLOPT_HEADERFUNCTION, [$this->headerParser, 'curlCallback']); |
120
|
|
|
curl_setopt($ch, CURLOPT_URL, $this->base . self::PATH); |
121
|
|
|
if (($this->contents = curl_exec($ch)) === false) { |
122
|
|
|
return false; |
123
|
|
|
} |
124
|
|
|
$this->time = time(); |
125
|
|
|
$this->statusCode = curl_getinfo($ch, CURLINFO_HTTP_CODE); // also works with FTP status codes |
126
|
|
|
$this->effectiveUri = curl_getinfo($ch, CURLINFO_EFFECTIVE_URL); |
127
|
|
|
curl_close($ch); |
128
|
|
|
$this->encoding = $this->headerParser->getCharset(); |
129
|
|
|
$this->maxAge = $this->headerParser->getMaxAge(); |
130
|
|
|
return true; |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
/** |
134
|
|
|
* Base uri |
135
|
|
|
* |
136
|
|
|
* @return string |
137
|
|
|
*/ |
138
|
|
|
public function getBaseUri() |
139
|
|
|
{ |
140
|
|
|
return $this->base; |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
/** |
144
|
|
|
* Effective uri |
145
|
|
|
* |
146
|
|
|
* @return string |
147
|
|
|
*/ |
148
|
|
|
public function getEffectiveUri() |
149
|
|
|
{ |
150
|
|
|
return $this->effectiveUri; |
151
|
|
|
} |
152
|
|
|
|
153
|
|
|
/** |
154
|
|
|
* Status code |
155
|
|
|
* |
156
|
|
|
* @return int|null |
157
|
|
|
*/ |
158
|
|
|
public function getStatusCode() |
159
|
|
|
{ |
160
|
|
|
return $this->statusCode; |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
/** |
164
|
|
|
* Body content |
165
|
|
|
* |
166
|
|
|
* @return string |
167
|
|
|
*/ |
168
|
|
|
public function getContents() |
169
|
|
|
{ |
170
|
|
|
return $this->contents; |
171
|
|
|
} |
172
|
|
|
|
173
|
|
|
/** |
174
|
|
|
* Encoding |
175
|
|
|
* |
176
|
|
|
* @return string |
177
|
|
|
*/ |
178
|
|
|
public function getEncoding() |
179
|
|
|
{ |
180
|
|
|
return $this->encoding; |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
/** |
184
|
|
|
* Next update timestamp |
185
|
|
|
* |
186
|
|
|
* @return int |
187
|
|
|
*/ |
188
|
|
|
public function nextUpdate() |
189
|
|
|
{ |
190
|
|
|
if ($this->statusCode === 503) { |
191
|
|
|
return $this->time + min(self::CACHE_TIME, $this->headerParser->getRetryAfter($this->time)); |
192
|
|
|
} |
193
|
|
|
return $this->time + self::CACHE_TIME; |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
/** |
197
|
|
|
* Valid until timestamp |
198
|
|
|
* |
199
|
|
|
* @return int |
200
|
|
|
*/ |
201
|
|
|
public function validUntil() |
202
|
|
|
{ |
203
|
|
|
return $this->time + max(self::CACHE_TIME, $this->maxAge); |
204
|
|
|
} |
205
|
|
|
} |
206
|
|
|
|