1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Sioen; |
4
|
|
|
|
5
|
|
|
use Sioen\HtmlToJson\Converter; |
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* Class HtmlToJson |
9
|
|
|
* |
10
|
|
|
* Converts html to a json object that can be understood by Sir Trevor |
11
|
|
|
* |
12
|
|
|
* @version 1.1.0 |
13
|
|
|
* @author Wouter Sioen <[email protected]> |
14
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php MIT |
15
|
|
|
*/ |
16
|
|
|
final class HtmlToJson |
17
|
|
|
{ |
18
|
|
|
/** @var array */ |
19
|
|
|
private $converters = array(); |
20
|
|
|
|
21
|
3 |
|
public function addConverter(Converter $converter) |
22
|
|
|
{ |
23
|
3 |
|
$this->converters[] = $converter; |
24
|
3 |
|
} |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* Converts html to the json Sir Trevor requires |
28
|
|
|
* |
29
|
|
|
* @param string $html |
30
|
|
|
* @return string The json string |
31
|
|
|
*/ |
32
|
3 |
|
public function toJson($html) |
33
|
|
|
{ |
34
|
|
|
// Strip white space between tags to prevent creation of empty #text nodes |
35
|
3 |
|
$html = preg_replace('~>\s+<~', '><', $html); |
36
|
3 |
|
$document = new \DOMDocument(); |
37
|
|
|
|
38
|
|
|
// Load UTF-8 HTML hack (from http://bit.ly/pVDyCt) |
39
|
3 |
|
$document->loadHTML('<?xml encoding="UTF-8">' . $html); |
40
|
3 |
|
$document->encoding = 'UTF-8'; |
41
|
|
|
|
42
|
|
|
// fetch the body of the document. All html is stored in there |
43
|
3 |
|
$body = $document->getElementsByTagName("body")->item(0); |
44
|
|
|
|
45
|
3 |
|
$data = array(); |
46
|
|
|
|
47
|
|
|
// loop trough the child nodes and convert them |
48
|
3 |
|
if ($body) { |
49
|
3 |
|
foreach ($body->childNodes as $node) { |
50
|
3 |
|
$data[] = $this->convert($node); |
51
|
3 |
|
} |
52
|
3 |
|
} |
53
|
|
|
|
54
|
3 |
|
return json_encode(array('data' => $data)); |
55
|
|
|
} |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* Converts one node to json |
59
|
|
|
* |
60
|
|
|
* @param \DOMElement $node |
61
|
|
|
* @return array |
62
|
|
|
*/ |
63
|
3 |
|
private function convert(\DOMElement $node) |
64
|
|
|
{ |
65
|
3 |
|
foreach ($this->converters as $converter) { |
66
|
3 |
|
if ($converter->matches($node)) { |
67
|
3 |
|
return $converter->toJson($node); |
68
|
|
|
} |
69
|
|
|
} |
70
|
|
|
} |
71
|
|
|
} |
72
|
|
|
|