|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Jclyons52\PagePreview; |
|
4
|
|
|
|
|
5
|
|
|
use Jclyons52\PHPQuery\Document; |
|
6
|
|
|
|
|
7
|
|
|
class Crawler |
|
8
|
|
|
{ |
|
9
|
|
|
private $document; |
|
10
|
|
|
|
|
11
|
39 |
|
public function __construct(Document $document) |
|
12
|
|
|
{ |
|
13
|
39 |
|
$this->document = $document; |
|
14
|
39 |
|
} |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* @return string |
|
18
|
|
|
*/ |
|
19
|
39 |
|
public function title() |
|
20
|
|
|
{ |
|
21
|
39 |
|
return $this->document->querySelector('title')->text(); |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @return mixed |
|
26
|
|
|
*/ |
|
27
|
39 |
|
public function metaKeywords() |
|
28
|
|
|
{ |
|
29
|
39 |
|
$keywordsElement = $this->document->querySelector('meta[name="keywords"]'); |
|
30
|
|
|
|
|
31
|
39 |
|
if (!$keywordsElement) { |
|
32
|
6 |
|
return []; |
|
33
|
2 |
|
} |
|
34
|
|
|
|
|
35
|
33 |
|
$keywordString = $keywordsElement->attr('content'); |
|
36
|
|
|
|
|
37
|
33 |
|
$keywords = explode(',', $keywordString); |
|
38
|
|
|
|
|
39
|
33 |
|
return array_map(function ($word) { |
|
40
|
33 |
|
return trim($word); |
|
41
|
|
|
|
|
42
|
33 |
|
}, $keywords); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* @param string $element |
|
47
|
|
|
* @return array |
|
48
|
|
|
*/ |
|
49
|
39 |
|
public function meta($element = null) |
|
50
|
|
|
{ |
|
51
|
39 |
|
$selector = "meta"; |
|
52
|
39 |
|
if ($element !== null) { |
|
53
|
39 |
|
$selector .= "[name='{$element}']"; |
|
54
|
39 |
|
$metaTags = $this->document->querySelector($selector); |
|
55
|
39 |
|
if ($metaTags === null) { |
|
56
|
6 |
|
return null; |
|
57
|
|
|
} |
|
58
|
33 |
|
return $metaTags->attr('content'); |
|
59
|
|
|
} |
|
60
|
39 |
|
$metaTags = $this->document->querySelectorAll($selector); |
|
61
|
|
|
|
|
62
|
39 |
|
return $this->metaTagsToArray($metaTags); |
|
63
|
2 |
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* get source attributes of all image tags on the page |
|
67
|
|
|
* @return array<String> |
|
68
|
|
|
*/ |
|
69
|
39 |
|
public function images() |
|
70
|
|
|
{ |
|
71
|
39 |
|
$images = $this->document->querySelectorAll('img'); |
|
72
|
|
|
|
|
73
|
39 |
|
if ($images === []) { |
|
74
|
3 |
|
return []; |
|
75
|
|
|
} |
|
76
|
36 |
|
return $images->attr('src'); |
|
77
|
|
|
|
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* @param \Jclyons52\PHPQuery\Support\NodeCollection $metaTags |
|
82
|
|
|
* @return array |
|
83
|
|
|
*/ |
|
84
|
39 |
|
private function metaTagsToArray($metaTags) |
|
85
|
|
|
{ |
|
86
|
39 |
|
$values = []; |
|
87
|
39 |
|
foreach ($metaTags as $meta) { |
|
88
|
39 |
|
$name = $meta->attr('name'); |
|
89
|
39 |
|
if ($name === '') { |
|
90
|
39 |
|
$name = $meta->attr('property'); |
|
91
|
26 |
|
} |
|
92
|
39 |
|
$content = $meta->attr('content'); |
|
93
|
39 |
|
if ($name === '' || $content == '') { |
|
94
|
39 |
|
continue; |
|
95
|
|
|
} |
|
96
|
33 |
|
$values[$name] = $content; |
|
97
|
26 |
|
} |
|
98
|
39 |
|
return $values; |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
|
|
|
jclyons52 /
page-preview
