|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace JeroenDesloovere\XmpMetadataExtractor; |
|
4
|
|
|
|
|
5
|
|
|
use DOMDocument; |
|
6
|
|
|
use JeroenDesloovere\XmpMetadataExtractor\Exception\FileNotFoundException; |
|
7
|
|
|
use SplFileInfo; |
|
8
|
|
|
|
|
9
|
|
|
final class XmpMetadataExtractor |
|
10
|
|
|
{ |
|
11
|
|
|
protected const RDF_ALT = 'rdf:Alt'; |
|
12
|
|
|
protected const RDF_BAG = 'rdf:Bag'; |
|
13
|
|
|
protected const RDF_LI = 'rdf:li'; |
|
14
|
|
|
protected const RDF_SEQ = 'rdf:Seq'; |
|
15
|
|
|
protected const POSSIBLE_CONTAINERS = [ |
|
16
|
|
|
self::RDF_ALT, |
|
17
|
|
|
self::RDF_BAG, |
|
18
|
|
|
self::RDF_SEQ, |
|
19
|
|
|
]; |
|
20
|
|
|
|
|
21
|
|
|
private function convertDomNode($node) |
|
22
|
|
|
{ |
|
23
|
|
|
switch ($node->nodeType) { |
|
24
|
|
|
case XML_CDATA_SECTION_NODE: |
|
25
|
|
|
case XML_TEXT_NODE: |
|
26
|
|
|
return trim($node->textContent); |
|
27
|
|
|
|
|
28
|
|
|
break; |
|
|
|
|
|
|
29
|
|
|
case XML_ELEMENT_NODE: |
|
30
|
|
|
return $this->convertXmlNode($node); |
|
31
|
|
|
|
|
32
|
|
|
break; |
|
33
|
|
|
} |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
private function convertXmlNode($node) |
|
37
|
|
|
{ |
|
38
|
|
|
$output = []; |
|
39
|
|
|
|
|
40
|
|
|
for ($i = 0, $m = $node->childNodes->length; $i < $m; $i++) { |
|
41
|
|
|
$child = $node->childNodes->item($i); |
|
42
|
|
|
$v = $this->convertDomNode($child); |
|
43
|
|
|
|
|
44
|
|
|
if (isset($child->tagName)) { |
|
45
|
|
|
$t = $child->tagName; |
|
46
|
|
|
if (!isset($output[$t])) { |
|
47
|
|
|
$output[$t] = array(); |
|
48
|
|
|
} |
|
49
|
|
|
$output[$t][] = $v; |
|
50
|
|
|
} elseif ($v || $v === '0') { |
|
51
|
|
|
$output = (string)$v; |
|
52
|
|
|
} |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
// Has attributes but isn't an array |
|
56
|
|
|
if ($node->attributes->length && !is_array($output)) { |
|
57
|
|
|
// Change output into an array. |
|
58
|
|
|
$output = array('@content' => $output); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
if (is_array($output)) { |
|
62
|
|
|
if ($node->attributes->length) { |
|
63
|
|
|
$a = array(); |
|
64
|
|
|
foreach ($node->attributes as $attrName => $attrNode) { |
|
65
|
|
|
$a[$attrName] = (string)$attrNode->value; |
|
66
|
|
|
} |
|
67
|
|
|
$output['@attributes'] = $a; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
foreach ($output as $t => $v) { |
|
71
|
|
|
// We are combining arrays for rdf:Bag, rdf:Alt, rdf:Seq |
|
72
|
|
|
if (in_array($t, self::POSSIBLE_CONTAINERS)) { |
|
73
|
|
|
if (!array_key_exists(self::RDF_LI, $v[0])) { |
|
74
|
|
|
break; |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
$output = $v[0][self::RDF_LI]; |
|
78
|
|
|
} elseif (is_array($v) && count($v) == 1 && $t != '@attributes') { |
|
79
|
|
|
$output[$t] = $v[0]; |
|
80
|
|
|
} |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
return $output; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
public function extractFromContent(string $content): array |
|
88
|
|
|
{ |
|
89
|
|
|
try { |
|
90
|
|
|
$doc = new DOMDocument(); |
|
|
|
|
|
|
91
|
|
|
$doc->loadXML($this->getXmpXmlString($content)); |
|
92
|
|
|
|
|
93
|
|
|
$root = $doc->documentElement; |
|
94
|
|
|
$output = $this->convertDomNode($root); |
|
95
|
|
|
$output['@root'] = $root->tagName; |
|
96
|
|
|
|
|
97
|
|
|
return $output; |
|
|
|
|
|
|
98
|
|
|
} catch (\Exception $e) { |
|
99
|
|
|
return []; |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
public function extractFromFile(string $file): array |
|
104
|
|
|
{ |
|
105
|
|
|
try { |
|
106
|
|
|
$file = new SplFileInfo($file); |
|
107
|
|
|
$contents = file_get_contents($file->getPathname()); |
|
108
|
|
|
} catch (\Exception $e) { |
|
109
|
|
|
throw new FileNotFoundException('The given File could not be found.'); |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
return $this->extractFromContent($contents); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
private function getXmpXmlString(string $content): string |
|
116
|
|
|
{ |
|
117
|
|
|
$xmpDataStart = strpos($content, '<x:xmpmeta'); |
|
118
|
|
|
$xmpDataEnd = strpos($content, '</x:xmpmeta>'); |
|
119
|
|
|
$xmpLength = $xmpDataEnd - $xmpDataStart; |
|
120
|
|
|
|
|
121
|
|
|
return substr($content, $xmpDataStart, $xmpLength + 12); |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|
jeroendesloovere /
xmp-metadata-extractor
Loading history...
The
breakstatement is not necessary if it is preceded for example by areturnstatement:If you would like to keep this construct to be consistent with other
casestatements, you can safely mark this issue as a false-positive.