|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Created by PhpStorm. |
|
4
|
|
|
* @author hashashiyyin [email protected] / [email protected] |
|
5
|
|
|
* Date: 22/04/24 |
|
6
|
|
|
* Time: 18:48 |
|
7
|
|
|
* |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
namespace Matecat\XmlParser; |
|
11
|
|
|
|
|
12
|
|
|
use DOMDocument; |
|
13
|
|
|
use Exception; |
|
14
|
|
|
use Matecat\XmlParser\Exception\InvalidXmlException; |
|
15
|
|
|
use Matecat\XmlParser\Exception\XmlParsingException; |
|
16
|
|
|
use RuntimeException; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* This class is copied from Symfony\Component\Config\Util\XmlUtils: |
|
20
|
|
|
* |
|
21
|
|
|
* Please see: |
|
22
|
|
|
* https://github.com/symfony/config/blob/v4.0.0/Util/XmlUtils.php |
|
23
|
|
|
*/ |
|
24
|
|
|
class XmlDomLoader { |
|
25
|
|
|
/** |
|
26
|
|
|
* Parses an XML string. |
|
27
|
|
|
* |
|
28
|
|
|
* @param string $content An XML string |
|
29
|
|
|
* @param Config|null $config |
|
30
|
|
|
* |
|
31
|
|
|
* @return DOMDocument |
|
32
|
|
|
* |
|
33
|
|
|
* @throws InvalidXmlException When parsing of XML with schema or callable produces any errors unrelated to the XML parsing itself |
|
34
|
|
|
* @throws XmlParsingException When parsing of XML file returns error |
|
35
|
|
|
*/ |
|
36
|
|
|
public static function load( $content, Config $config = null ) { |
|
37
|
|
|
if ( !extension_loaded( 'dom' ) ) { |
|
38
|
|
|
throw new RuntimeException( 'Extension DOM is required.' ); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
if ( is_null( $config ) ) { |
|
42
|
|
|
$config = new Config(); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
$internalErrors = libxml_use_internal_errors( true ); |
|
46
|
|
|
$disableEntities = libxml_disable_entity_loader(); |
|
47
|
|
|
libxml_clear_errors(); |
|
48
|
|
|
|
|
49
|
|
|
$dom = new DOMDocument( '1.0', 'UTF-8' ); |
|
50
|
|
|
$dom->validateOnParse = true; |
|
51
|
|
|
|
|
52
|
|
|
if ( is_string( $config->getSetRootElement() ) && !empty( $config->getSetRootElement() ) ) { |
|
53
|
|
|
$content = "<{$config->getSetRootElement()}>$content</{$config->getSetRootElement()}>"; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
$res = $dom->loadXML( $content, $config->getXML_OPTIONS() ); |
|
57
|
|
|
|
|
58
|
|
|
if ( !$res ) { |
|
59
|
|
|
libxml_disable_entity_loader( $disableEntities ); |
|
60
|
|
|
|
|
61
|
|
|
throw new XmlParsingException( implode( "\n", static::getXmlErrors( $internalErrors ) ) ); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
$dom->normalizeDocument(); |
|
65
|
|
|
|
|
66
|
|
|
libxml_use_internal_errors( $internalErrors ); |
|
67
|
|
|
libxml_disable_entity_loader( $disableEntities ); |
|
68
|
|
|
|
|
69
|
|
|
foreach ( $dom->childNodes as $child ) { |
|
70
|
|
|
if ( XML_DOCUMENT_TYPE_NODE === $child->nodeType && !$config->getAllowDocumentType() ) { |
|
71
|
|
|
throw new XmlParsingException( 'Document types are not allowed.' ); |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
if ( null !== $config->getSchemaOrCallable() ) { |
|
76
|
|
|
$internalErrors = libxml_use_internal_errors( true ); |
|
77
|
|
|
libxml_clear_errors(); |
|
78
|
|
|
|
|
79
|
|
|
$e = null; |
|
80
|
|
|
if ( is_callable( $config->getSchemaOrCallable() ) ) { |
|
81
|
|
|
try { |
|
82
|
|
|
$valid = call_user_func( $config->getSchemaOrCallable(), $dom, $internalErrors ); |
|
83
|
|
|
} catch ( Exception $e ) { |
|
84
|
|
|
$valid = false; |
|
85
|
|
|
} |
|
86
|
|
|
} elseif ( !is_array( $config->getSchemaOrCallable() ) && is_file( (string)$config->getSchemaOrCallable() ) ) { |
|
87
|
|
|
$schemaSource = file_get_contents( (string)$config->getSchemaOrCallable() ); |
|
88
|
|
|
$valid = @$dom->schemaValidateSource( $schemaSource ); |
|
89
|
|
|
} else { |
|
90
|
|
|
libxml_use_internal_errors( $internalErrors ); |
|
91
|
|
|
|
|
92
|
|
|
throw new XmlParsingException( 'The schemaOrCallable argument has to be a valid path to XSD file or callable.' ); |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
if ( !$valid ) { |
|
96
|
|
|
$messages = static::getXmlErrors( $internalErrors ); |
|
97
|
|
|
if ( empty( $messages ) ) { |
|
98
|
|
|
throw new InvalidXmlException( 'The XML is not valid.', 0, $e ); |
|
99
|
|
|
} |
|
100
|
|
|
throw new XmlParsingException( implode( "\n", $messages ), 0, $e ); |
|
101
|
|
|
} |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
libxml_clear_errors(); |
|
105
|
|
|
libxml_use_internal_errors( $internalErrors ); |
|
106
|
|
|
|
|
107
|
|
|
return $dom; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* @param $internalErrors |
|
112
|
|
|
* |
|
113
|
|
|
* @return array |
|
114
|
|
|
*/ |
|
115
|
|
|
private static function getXmlErrors( $internalErrors ) { |
|
116
|
|
|
$errors = []; |
|
117
|
|
|
foreach ( libxml_get_errors() as $error ) { |
|
118
|
|
|
$errors[] = sprintf( |
|
119
|
|
|
'[%s %s] %s (in %s - line %d, column %d)', |
|
120
|
|
|
LIBXML_ERR_WARNING == $error->level ? 'WARNING' : 'ERROR', |
|
121
|
|
|
$error->code, |
|
122
|
|
|
trim( $error->message ), |
|
123
|
|
|
$error->file ?: 'n/a', |
|
124
|
|
|
$error->line, |
|
125
|
|
|
$error->column |
|
126
|
|
|
); |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
libxml_clear_errors(); |
|
130
|
|
|
libxml_use_internal_errors( $internalErrors ); |
|
131
|
|
|
|
|
132
|
|
|
return $errors; |
|
133
|
|
|
} |
|
134
|
|
|
} |
matecat /
xml-dom-parser
