1
|
|
|
<?php |
2
|
|
|
namespace vipnytt\RobotsTxtParser\Parser; |
3
|
|
|
|
4
|
|
|
use vipnytt\RobotsTxtParser\Exceptions\ClientException; |
5
|
|
|
|
6
|
|
|
/** |
7
|
|
|
* Trait UrlParser |
8
|
|
|
* |
9
|
|
|
* @package vipnytt\RobotsTxtParser\Parser |
10
|
|
|
*/ |
11
|
|
|
trait UrlParser |
12
|
|
|
{ |
13
|
|
|
/** |
14
|
|
|
* Convert relative to full URL |
15
|
|
|
* |
16
|
|
|
* @param string $url |
17
|
|
|
* @param string $base |
18
|
|
|
* @return string |
19
|
|
|
* @throws ClientException |
20
|
|
|
*/ |
21
|
|
|
private function urlConvertToFull($url, $base) |
22
|
|
|
{ |
23
|
|
|
$url = $this->urlEncode($url); |
24
|
|
|
if ($this->urlValidate($url)) { |
25
|
|
|
return $url; |
26
|
|
|
} elseif (mb_stripos($url, '/') === 0) { |
27
|
|
|
return $this->urlBase($base) . $url; |
28
|
|
|
} |
29
|
|
|
throw new ClientException('Invalid URL'); |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* URL encoder according to RFC 3986 |
34
|
|
|
* Returns a string containing the encoded URL with disallowed characters converted to their percentage encodings. |
35
|
|
|
* @link http://publicmind.in/blog/url-encoding/ |
36
|
|
|
* |
37
|
|
|
* @param string $url |
38
|
|
|
* @return string |
39
|
|
|
*/ |
40
|
|
|
protected function urlEncode($url) |
41
|
|
|
{ |
42
|
|
|
$reserved = [ |
43
|
|
|
":" => '!%3A!ui', |
44
|
|
|
"/" => '!%2F!ui', |
45
|
|
|
"?" => '!%3F!ui', |
46
|
|
|
"#" => '!%23!ui', |
47
|
|
|
"[" => '!%5B!ui', |
48
|
|
|
"]" => '!%5D!ui', |
49
|
|
|
"@" => '!%40!ui', |
50
|
|
|
"!" => '!%21!ui', |
51
|
|
|
"$" => '!%24!ui', |
52
|
|
|
"&" => '!%26!ui', |
53
|
|
|
"'" => '!%27!ui', |
54
|
|
|
"(" => '!%28!ui', |
55
|
|
|
")" => '!%29!ui', |
56
|
|
|
"*" => '!%2A!ui', |
57
|
|
|
"+" => '!%2B!ui', |
58
|
|
|
"," => '!%2C!ui', |
59
|
|
|
";" => '!%3B!ui', |
60
|
|
|
"=" => '!%3D!ui', |
61
|
|
|
"%" => '!%25!ui' |
62
|
|
|
]; |
63
|
|
|
return preg_replace(array_values($reserved), array_keys($reserved), rawurlencode($url)); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* Validate URL |
68
|
|
|
* |
69
|
|
|
* @param string $url |
70
|
|
|
* @return bool |
71
|
|
|
*/ |
72
|
|
|
private function urlValidate($url) |
73
|
|
|
{ |
74
|
|
|
return ( |
75
|
|
|
filter_var($url, FILTER_VALIDATE_URL) && |
76
|
|
|
($parsed = parse_url($url)) !== false && |
77
|
|
|
$this->urlValidateHost($parsed['host']) && |
78
|
|
|
$this->urlValidateScheme($parsed['scheme']) |
79
|
|
|
); |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* Validate host name |
84
|
|
|
* |
85
|
|
|
* @link http://stackoverflow.com/questions/1755144/how-to-validate-domain-name-in-php |
86
|
|
|
* |
87
|
|
|
* @param string $host |
88
|
|
|
* @return bool |
89
|
|
|
*/ |
90
|
|
|
private static function urlValidateHost($host) |
91
|
|
|
{ |
92
|
|
|
return ( |
93
|
|
|
preg_match("/^([a-z\d](-*[a-z\d])*)(\.([a-z\d](-*[a-z\d])*))*$/i", $host) //valid chars check |
94
|
|
|
&& preg_match("/^.{1,253}$/", $host) //overall length check |
95
|
|
|
&& preg_match("/^[^\.]{1,63}(\.[^\.]{1,63})*$/", $host) //length of each label |
96
|
|
|
&& !filter_var($host, FILTER_VALIDATE_IP) //is not an IP address |
97
|
|
|
); |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
/** |
101
|
|
|
* Validate URL scheme |
102
|
|
|
* |
103
|
|
|
* @param string $scheme |
104
|
|
|
* @return bool |
105
|
|
|
*/ |
106
|
|
|
private static function urlValidateScheme($scheme) |
107
|
|
|
{ |
108
|
|
|
return in_array(mb_strtolower($scheme), [ |
109
|
|
|
'http', |
110
|
|
|
'https', |
111
|
|
|
'ftp', |
112
|
|
|
'sftp', |
113
|
|
|
'ftps', |
114
|
|
|
] |
115
|
|
|
); |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
/** |
119
|
|
|
* Base URL |
120
|
|
|
* |
121
|
|
|
* @param string $url |
122
|
|
|
* @return string |
123
|
|
|
* @throws ClientException |
124
|
|
|
*/ |
125
|
|
|
protected function urlBase($url) |
126
|
|
|
{ |
127
|
|
|
if ($this->urlValidate($url) === false) { |
128
|
|
|
throw new ClientException('Invalid URL'); |
129
|
|
|
} |
130
|
|
|
$parts = [ |
131
|
|
|
'scheme' => parse_url($url, PHP_URL_SCHEME), |
132
|
|
|
'host' => parse_url($url, PHP_URL_HOST), |
133
|
|
|
]; |
134
|
|
|
$parts['port'] = is_int($port = parse_url($url, PHP_URL_PORT)) ? $port : getservbyname($parts['scheme'], 'tcp'); |
135
|
|
|
return $parts['scheme'] . '://' . $parts['host'] . ':' . $parts['port']; |
136
|
|
|
} |
137
|
|
|
} |
138
|
|
|
|