|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of Nkey/LinkCheck package |
|
4
|
|
|
* |
|
5
|
|
|
* @package Nkey\LinkCheck |
|
6
|
|
|
*/ |
|
7
|
|
|
namespace Nkey\LinkCheck; |
|
8
|
|
|
|
|
9
|
|
|
use Generics\Client\HttpClientFactory; |
|
10
|
|
|
use Generics\Socket\Url; |
|
11
|
|
|
use Generics\Streams\StandardOutputStream; |
|
12
|
|
|
use Generics\Streams\HttpStream; |
|
13
|
|
|
use Generics\Util\UrlParser; |
|
14
|
|
|
use Generics\Streams\OutputStream; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* This class provides the link checking functionality. |
|
18
|
|
|
* |
|
19
|
|
|
* @author Maik Greubel <[email protected]> |
|
20
|
|
|
*/ |
|
21
|
|
|
class LinkCheckProvider |
|
22
|
|
|
{ |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* |
|
26
|
|
|
* @var HttpStream |
|
27
|
|
|
*/ |
|
28
|
|
|
private $httpClient; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* |
|
32
|
|
|
* @var OutputStream |
|
33
|
|
|
*/ |
|
34
|
|
|
private $output; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* |
|
38
|
|
|
* @var Url |
|
39
|
|
|
*/ |
|
40
|
|
|
private $url; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* Create a new instance of LinkCheckProvider |
|
44
|
|
|
* |
|
45
|
|
|
* @param Url $url |
|
46
|
|
|
* The url to check |
|
47
|
|
|
*/ |
|
48
|
1 |
|
public function __construct(Url $url) |
|
49
|
|
|
{ |
|
50
|
1 |
|
$this->httpClient = HttpClientFactory::get($url); |
|
51
|
1 |
|
$this->output = new StandardOutputStream(); |
|
52
|
1 |
|
$this->url = $url; |
|
53
|
1 |
|
} |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* Override the standard output stream using a custom output stream instance |
|
57
|
|
|
* |
|
58
|
|
|
* @param OutputStream $stream |
|
59
|
|
|
*/ |
|
60
|
1 |
|
public function setOutput(OutputStream $stream) |
|
61
|
|
|
{ |
|
62
|
1 |
|
$this->output = $stream; |
|
63
|
1 |
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* Perform check |
|
67
|
|
|
* - Retrieve the contents of url |
|
68
|
|
|
* - Extract all anchor links |
|
69
|
|
|
* - Check all anchor links for existance (response code < 400) |
|
70
|
|
|
* |
|
71
|
|
|
* @param array $options |
|
72
|
|
|
* The options for check ('recursive' => perform a recursive check of all same-site-links) |
|
73
|
|
|
*/ |
|
74
|
1 |
|
public function check(array $options = array()) |
|
75
|
|
|
{ |
|
76
|
1 |
|
$this->httpClient->request('GET'); |
|
77
|
|
|
|
|
78
|
1 |
|
if ($this->httpClient->getResponseCode() != 200) { |
|
79
|
|
|
$this->output->write("Invalid response code " . $this->httpClient->getResponseCode()); |
|
80
|
|
|
return; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
1 |
|
$response = ""; |
|
84
|
1 |
|
$size = $this->httpClient->getPayload()->count(); |
|
85
|
1 |
|
if (isset($this->httpClient->getHeaders()['Content-Length'])) { |
|
86
|
1 |
|
$size = $this->httpClient->getHeaders()['Content-Length']; |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
1 |
|
while ($this->httpClient->getPayload()->ready()) { |
|
90
|
1 |
|
$response .= $this->httpClient->getPayload()->read($size); |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
1 |
|
$this->extractAndCheckUrls($response, $options); |
|
94
|
1 |
|
} |
|
95
|
|
|
|
|
96
|
1 |
|
private function extractAndCheckUrls(string $response, array $options) |
|
97
|
|
|
{ |
|
98
|
1 |
|
$matches = []; |
|
99
|
|
|
|
|
100
|
1 |
|
if (preg_match_all('#<a href="([^\"]*)">#', $response, $matches)) { |
|
101
|
1 |
|
array_shift($matches); |
|
102
|
1 |
|
$matches = $matches[0]; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
1 |
|
if (count($matches) == 0) { |
|
106
|
|
|
return; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
1 |
|
foreach ($matches as $match) { |
|
110
|
1 |
|
if (substr($match, 0, 4) != 'http') { |
|
111
|
1 |
|
$match = sprintf("%s://%s:%d%s", $this->url->getScheme(), $this->url->getAddress(), $this->url->getPort(), $match); |
|
112
|
|
|
} |
|
113
|
1 |
|
$url = UrlParser::parseUrl($match); |
|
114
|
|
|
|
|
115
|
1 |
|
$this->output->write($url); |
|
116
|
1 |
|
if ($url->getAddress() == $this->url->getAddress()) { |
|
117
|
1 |
|
$this->checkSameSiteurl($url, $options); |
|
118
|
|
|
} |
|
119
|
1 |
|
$this->output->write("\n"); |
|
120
|
|
|
} |
|
121
|
1 |
|
} |
|
122
|
|
|
|
|
123
|
1 |
|
private function checkSameSiteUrl(Url $url, array $options) |
|
124
|
|
|
{ |
|
125
|
1 |
|
if (isset($options['recursive']) && $options['recursive'] == true) { |
|
126
|
|
|
$subProvider = new LinkCheckProvider($url); |
|
127
|
|
|
$subProvider->check($options); |
|
128
|
|
|
} else { |
|
129
|
1 |
|
$http = HttpClientFactory::get($url); |
|
130
|
1 |
|
$http->setTimeout(10); |
|
131
|
1 |
|
$http->request('HEAD'); |
|
132
|
1 |
|
if ($http->getResponseCode() < 400) { |
|
133
|
1 |
|
$this->output->write(" OK"); |
|
134
|
|
|
} |
|
135
|
|
|
} |
|
136
|
|
|
} |
|
137
|
|
|
} |
maikgreubel /
link-check
