|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace YusufKandemir\MicrodataParser; |
|
4
|
|
|
|
|
5
|
|
|
class MicrodataParser |
|
6
|
|
|
{ |
|
7
|
|
|
/** @var callable|null */ |
|
8
|
|
|
protected $absoluteUriHandler; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* @param callable|null $absoluteUriHandler |
|
12
|
|
|
* |
|
13
|
|
|
* @see MicrodataElementParser::$absoluteUriHandler |
|
14
|
|
|
*/ |
|
15
|
39 |
|
public function __construct(callable $absoluteUriHandler = null) |
|
16
|
|
|
{ |
|
17
|
39 |
|
$this->absoluteUriHandler = $absoluteUriHandler; |
|
18
|
39 |
|
} |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Parses HTML string, extracts microdata from it |
|
22
|
|
|
* |
|
23
|
|
|
* @param string $html HTML string to be parsed |
|
24
|
|
|
* @param string $documentURI DocumentURI to be used in absolutizing URIs |
|
25
|
|
|
* |
|
26
|
|
|
* @return \stdClass |
|
27
|
|
|
*/ |
|
28
|
15 |
|
public function parseHTML(string $html, string $documentURI = '') : \stdClass |
|
29
|
|
|
{ |
|
30
|
15 |
|
$dom = new \DOMDocument; |
|
31
|
15 |
|
$dom->loadHTML($html, \LIBXML_NOERROR); |
|
32
|
15 |
|
$dom->documentURI = $documentURI; |
|
33
|
|
|
|
|
34
|
15 |
|
return $this->parse($dom); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* Parses HTML file, extracts microdata from it |
|
39
|
|
|
* |
|
40
|
|
|
* @param string $path Path to the file to be parsed |
|
41
|
|
|
* @param string $documentURI DocumentURI to be used in absolutizing URIs |
|
42
|
|
|
* |
|
43
|
|
|
* @return \stdClass |
|
44
|
|
|
*/ |
|
45
|
12 |
|
public function parseHTMLFile(string $path, string $documentURI = '') : \stdClass |
|
46
|
|
|
{ |
|
47
|
12 |
|
$dom = new \DOMDocument; |
|
48
|
12 |
|
$dom->loadHTMLFile($path, \LIBXML_NOERROR); |
|
49
|
12 |
|
$dom->documentURI = $documentURI; |
|
50
|
|
|
|
|
51
|
12 |
|
return $this->parse($dom); |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* Creates a MicrodataParser from a DOMDocument instance. |
|
56
|
|
|
* If you have MicrodataDOMDocument then instantiate MicrodataParser class directly to avoid conversion. |
|
57
|
|
|
* |
|
58
|
|
|
* @param \DOMDocument $domDocument DOMDocument to be parsed. |
|
59
|
|
|
* @param string $documentURI If non-empty value is provided, |
|
60
|
|
|
* it will be new value of documentURI property of $domDocument. |
|
61
|
|
|
* |
|
62
|
|
|
* @return \stdClass |
|
63
|
|
|
*/ |
|
64
|
12 |
|
public function parseDOMDocument(\DOMDocument $domDocument, string $documentURI = '') : \stdClass |
|
65
|
|
|
{ |
|
66
|
12 |
|
if (!empty($documentURI)) { |
|
67
|
|
|
$domDocument->documentURI = $documentURI; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
12 |
|
return $this->parse($domDocument); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* @param \DOMDocument $dom |
|
75
|
|
|
* |
|
76
|
|
|
* @see MicrodataElementParser::$absoluteUriHandler |
|
77
|
|
|
* |
|
78
|
|
|
* @return \stdClass |
|
79
|
|
|
*/ |
|
80
|
39 |
|
protected function parse(\DOMDocument $dom) : \stdClass |
|
81
|
|
|
{ |
|
82
|
39 |
|
$elementParser = new MicrodataElementParser($this->absoluteUriHandler); |
|
83
|
39 |
|
$documentParser = new MicrodataDocumentParser($dom, $elementParser); |
|
84
|
|
|
|
|
85
|
39 |
|
return $documentParser->parse(); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* @param callable|null $absoluteUriHandler |
|
90
|
|
|
* |
|
91
|
|
|
* @see MicrodataElementParser::$absoluteUriHandler |
|
92
|
|
|
* |
|
93
|
|
|
* @return MicrodataParser |
|
94
|
|
|
*/ |
|
95
|
3 |
|
public function setAbsoluteUriHandler(callable $absoluteUriHandler = null) : self |
|
96
|
|
|
{ |
|
97
|
3 |
|
$this->absoluteUriHandler = $absoluteUriHandler; |
|
98
|
|
|
|
|
99
|
3 |
|
return $this; |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
|
yusufkandemir /
microdata-parser
