|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace SP\Spiderling; |
|
4
|
|
|
|
|
5
|
|
|
use GuzzleHttp\Psr7\Uri; |
|
6
|
|
|
use Psr\Http\Message\UriInterface; |
|
7
|
|
|
use DOMDocument; |
|
8
|
|
|
use DOMXPath; |
|
9
|
|
|
use InvalidArgumentException; |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* @author Ivan Kerin <[email protected]> |
|
13
|
|
|
* @copyright 2015, Clippings Ltd. |
|
14
|
|
|
* @license http://spdx.org/licenses/BSD-3-Clause |
|
15
|
|
|
*/ |
|
16
|
|
|
class Html |
|
17
|
|
|
{ |
|
18
|
|
|
/** |
|
19
|
|
|
* @var DOMDocument |
|
20
|
|
|
*/ |
|
21
|
|
|
private $document; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @var DOMXPath |
|
25
|
|
|
*/ |
|
26
|
|
|
private $xpath; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @param string $html |
|
30
|
|
|
*/ |
|
31
|
1 |
|
public function __construct($html) |
|
32
|
|
|
{ |
|
33
|
|
|
$this->document = new DOMDocument(); |
|
34
|
|
|
$this->document->loadHtml($html); |
|
35
|
|
|
$this->xpath = new DOMXPath($this->document); |
|
36
|
1 |
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @param string $attribute |
|
40
|
|
|
* @param UriInterface $base |
|
41
|
|
|
*/ |
|
42
|
|
|
private function resolveLinkAttribute($attribute, UriInterface $base) |
|
43
|
|
|
{ |
|
44
|
|
|
$elements = $this->xpath->query( |
|
45
|
|
|
"//*[@$attribute and not(contains(@$attribute, \"://\"))]" |
|
46
|
|
|
); |
|
47
|
|
|
|
|
48
|
|
|
foreach ($elements as $element) { |
|
49
|
|
|
try { |
|
50
|
|
|
$resolved = Uri::resolve($base, $element->getAttribute($attribute)); |
|
51
|
|
|
$element->setAttribute($attribute, $resolved->__toString()); |
|
52
|
|
|
} catch (InvalidArgumentException $e) { |
|
53
|
|
|
// Tolerate invalid urls |
|
54
|
|
|
} |
|
55
|
|
|
} |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* Add a prefix to all relative links (src, href and action) |
|
60
|
|
|
* |
|
61
|
|
|
* @param UriInterface $base |
|
62
|
|
|
*/ |
|
63
|
1 |
|
public function resolveLinks(UriInterface $base) |
|
64
|
|
|
{ |
|
65
|
|
|
$this->resolveLinkAttribute('href', $base); |
|
66
|
|
|
$this->resolveLinkAttribute('src', $base); |
|
67
|
|
|
$this->resolveLinkAttribute('action', $base); |
|
68
|
|
|
|
|
69
|
1 |
|
return $this; |
|
70
|
1 |
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* @return string |
|
74
|
|
|
*/ |
|
75
|
|
|
public function get() |
|
76
|
|
|
{ |
|
77
|
|
|
return $this->document->saveHtml(); |
|
78
|
|
|
} |
|
79
|
|
|
} |
|
80
|
|
|
|
spiderling-php /
spiderling
