1
|
|
|
<?php |
2
|
|
|
namespace SEOCheckup; |
3
|
|
|
|
4
|
|
|
class Helpers |
5
|
|
|
{ |
6
|
|
|
/** |
7
|
|
|
* @var array $data |
8
|
|
|
*/ |
9
|
|
|
private $data; |
10
|
|
|
|
11
|
|
|
/** |
12
|
|
|
* @var array $links |
13
|
|
|
*/ |
14
|
|
|
private $links; |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* Helpers constructor |
18
|
|
|
* @param array $data |
19
|
|
|
*/ |
20
|
|
|
public function __construct($data) |
21
|
|
|
{ |
22
|
|
|
$this->data = $data; |
23
|
|
|
} |
24
|
|
|
|
25
|
|
|
/** |
26
|
|
|
* Get links in a page |
27
|
|
|
* |
28
|
|
|
* @param \DOMDocument $dom |
29
|
|
|
* @return $this |
30
|
|
|
*/ |
31
|
|
|
public function Links($dom) |
32
|
|
|
{ |
33
|
|
|
$tags = $dom->getElementsByTagName('a'); |
34
|
|
|
$links = array(); |
35
|
|
|
|
36
|
|
|
if($tags->length) |
37
|
|
|
{ |
38
|
|
|
foreach($tags as $item) |
39
|
|
|
{ |
40
|
|
|
$link = $item->getAttribute('href'); |
41
|
|
|
|
42
|
|
|
if($link != '' && strpos($link,'#') !== 0 && strpos(strtolower($link),'javascript:') !== 0) |
43
|
|
|
{ |
44
|
|
|
$link = parse_url($link); |
45
|
|
|
|
46
|
|
|
if(!isset($link['scheme'])) |
47
|
|
|
{ |
48
|
|
|
$link['scheme'] = $this->data['parsed_url']['scheme']; |
49
|
|
|
} |
50
|
|
|
|
51
|
|
|
if(!isset($link['host'])) |
52
|
|
|
{ |
53
|
|
|
$link['host'] = $this->data['parsed_url']['host']; |
54
|
|
|
} |
55
|
|
|
|
56
|
|
|
if(!isset($link['path'])) |
57
|
|
|
{ |
58
|
|
|
$link['path'] = ''; |
59
|
|
|
} else { |
60
|
|
|
if(strpos($link['path'],'/') !== 0) |
61
|
|
|
{ |
62
|
|
|
$link['path'] = '/'.$link['path']; |
63
|
|
|
} |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
if(!isset($link['query'])) |
67
|
|
|
{ |
68
|
|
|
$link['query'] = ''; |
69
|
|
|
} else { |
70
|
|
|
$link['query'] = '?'.$link['query']; |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
$links[] = $link['scheme'].'://'.$link['host'].$link['path'].$link['query']; |
74
|
|
|
} |
75
|
|
|
} |
76
|
|
|
} |
77
|
|
|
|
78
|
|
|
$this->links = array_unique($links); |
79
|
|
|
|
80
|
|
|
return $this; |
81
|
|
|
} |
82
|
|
|
|
83
|
|
|
/** |
84
|
|
|
* Return page links |
85
|
|
|
* |
86
|
|
|
* @return array |
87
|
|
|
*/ |
88
|
|
|
public function GetLinks() |
89
|
|
|
{ |
90
|
|
|
return $this->links; |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* Get link attributes in a page |
95
|
|
|
* |
96
|
|
|
* @param \DOMDocument $dom |
97
|
|
|
* @param string $tag |
98
|
|
|
* @param string $attr |
99
|
|
|
* @return array |
100
|
|
|
*/ |
101
|
|
|
public function GetAttributes($dom, $tag = 'a', $attr = 'href') |
102
|
|
|
{ |
103
|
|
|
$tags = $dom->getElementsByTagName($tag); |
104
|
|
|
$links = array(); |
105
|
|
|
|
106
|
|
|
if($tags->length) |
107
|
|
|
{ |
108
|
|
|
foreach($tags as $item) |
109
|
|
|
{ |
110
|
|
|
$links[] = $item->getAttribute($attr); |
111
|
|
|
} |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
return array_unique($links); |
115
|
|
|
} |
116
|
|
|
|
117
|
|
|
/** |
118
|
|
|
* Whitespace cleaner |
119
|
|
|
* |
120
|
|
|
* @param $input |
121
|
|
|
* @return null|string|string[] |
122
|
|
|
*/ |
123
|
|
|
public function Whitespace($input) |
124
|
|
|
{ |
125
|
|
|
return preg_replace('!\s+!', ' ', $input); |
126
|
|
|
} |
127
|
|
|
} |