|
1
|
|
|
<?php |
|
2
|
|
|
namespace XmlSchemaValidator; |
|
3
|
|
|
|
|
4
|
|
|
use DOMDocument; |
|
5
|
|
|
use DOMXPath; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* This class is an XML schema validator |
|
9
|
|
|
* It is needed because some XML can contain more than one external schema |
|
10
|
|
|
* and DOM library fails to load it. |
|
11
|
|
|
*/ |
|
12
|
|
|
class SchemaValidator |
|
13
|
|
|
{ |
|
14
|
|
|
/** @var DOMDocument */ |
|
15
|
|
|
private $document; |
|
16
|
|
|
|
|
17
|
|
|
/** @var string */ |
|
18
|
|
|
private $error = ''; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* SchemaValidator constructor. |
|
22
|
|
|
* |
|
23
|
|
|
* @param string $content |
|
24
|
|
|
* @throws \InvalidArgumentException if content is empty |
|
25
|
|
|
* @throws SchemaValidatorException if malformed xml content |
|
26
|
|
|
*/ |
|
27
|
13 |
|
public function __construct(string $content) |
|
28
|
|
|
{ |
|
29
|
13 |
|
if ('' === $content) { |
|
30
|
1 |
|
throw new \InvalidArgumentException('The content to validate must be a non-empty string'); |
|
31
|
|
|
} |
|
32
|
12 |
|
$document = new DOMDocument(); |
|
33
|
|
|
try { |
|
34
|
|
|
LibXmlException::useInternalErrors(function () use ($content, $document) { |
|
35
|
12 |
|
$document->loadXML($content, LIBXML_NOWARNING); |
|
36
|
12 |
|
}); |
|
37
|
1 |
|
} catch (LibXmlException $ex) { |
|
38
|
1 |
|
throw new SchemaValidatorException('Malformed XML Document: ' . $ex->getMessage()); |
|
39
|
|
|
} |
|
40
|
11 |
|
$this->document = $document; |
|
41
|
11 |
|
} |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* Validate the content by: |
|
45
|
|
|
* - Create the Schemas collection from the document |
|
46
|
|
|
* - Validate using validateWithSchemas |
|
47
|
|
|
* |
|
48
|
|
|
* @see validateWithSchemas |
|
49
|
|
|
* @return bool |
|
50
|
|
|
*/ |
|
51
|
8 |
|
public function validate(): bool |
|
52
|
|
|
{ |
|
53
|
8 |
|
$this->error = ''; |
|
54
|
|
|
try { |
|
55
|
|
|
// create the schemas collection |
|
56
|
8 |
|
$schemas = $this->buildSchemas(); |
|
57
|
|
|
// validate the document against the schema collection |
|
58
|
7 |
|
$this->validateWithSchemas($schemas); |
|
|
|
|
|
|
59
|
4 |
|
} catch (LibXmlException $ex) { |
|
60
|
3 |
|
$this->error = $ex->getMessage(); |
|
61
|
3 |
|
return false; |
|
62
|
|
|
} |
|
63
|
4 |
|
return true; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
4 |
|
public function getLastError(): string |
|
67
|
|
|
{ |
|
68
|
4 |
|
return $this->error; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Validate against a list of schemas (if any) |
|
73
|
|
|
* |
|
74
|
|
|
* @param Schemas $schemas |
|
75
|
|
|
* @return void |
|
76
|
|
|
* |
|
77
|
|
|
* @throws LibXmlException if schema validation fails |
|
78
|
|
|
*/ |
|
79
|
10 |
|
public function validateWithSchemas(Schemas $schemas) |
|
80
|
|
|
{ |
|
81
|
|
|
// create the schemas collection, then validate the document against the schemas |
|
82
|
10 |
|
if (! $schemas->count()) { |
|
83
|
2 |
|
return; |
|
84
|
|
|
} |
|
85
|
|
|
// build the unique importing schema |
|
86
|
8 |
|
$xsd = $schemas->getImporterXsd(); |
|
87
|
8 |
|
$document = $this->document; |
|
88
|
|
|
LibXmlException::useInternalErrors(function () use ($document, $xsd) { |
|
89
|
8 |
|
$document->schemaValidateSource($xsd); |
|
90
|
8 |
|
}); |
|
91
|
4 |
|
} |
|
92
|
|
|
|
|
93
|
|
|
/** |
|
94
|
|
|
* Retrieve a list of namespaces based on the schemaLocation attributes |
|
95
|
|
|
* |
|
96
|
|
|
* @throws SchemaValidatorException if the content of schemaLocation is not an even number of uris |
|
97
|
|
|
* @return Schemas|Schema[] |
|
98
|
|
|
*/ |
|
99
|
8 |
|
public function buildSchemas(): Schemas |
|
100
|
|
|
{ |
|
101
|
8 |
|
$schemas = new Schemas(); |
|
102
|
8 |
|
$xpath = new DOMXPath($this->document); |
|
103
|
|
|
|
|
104
|
|
|
// get the http://www.w3.org/2001/XMLSchema-instance namespace (it could not be 'xsi') |
|
105
|
8 |
|
$xsi = $this->document->lookupPrefix('http://www.w3.org/2001/XMLSchema-instance'); |
|
106
|
|
|
|
|
107
|
|
|
// the namespace is not registered, no need to continue |
|
108
|
8 |
|
if (! $xsi) { |
|
109
|
1 |
|
return $schemas; |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
// get all the xsi:schemaLocation attributes in the document |
|
113
|
|
|
/** @var \DOMNodeList|false $schemasList */ |
|
114
|
7 |
|
$schemasList = $xpath->query("//@$xsi:schemaLocation"); |
|
115
|
|
|
|
|
116
|
|
|
// schemaLocation attribute not found, no need to continue |
|
117
|
7 |
|
if (false === $schemasList || 0 === $schemasList->length) { |
|
118
|
1 |
|
return $schemas; |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
// process every schemaLocation for even parts |
|
122
|
6 |
|
for ($s = 0; $s < $schemasList->length; $s++) { |
|
123
|
6 |
|
$node = $schemasList->item($s); |
|
124
|
6 |
|
if (null === $node) { |
|
125
|
|
|
continue; |
|
126
|
|
|
} |
|
127
|
|
|
// get the node content |
|
128
|
6 |
|
$content = $node->nodeValue; |
|
129
|
|
|
// get parts without inner spaces |
|
130
|
6 |
|
$parts = array_values(array_filter(explode(' ', $content))); |
|
131
|
6 |
|
$partsCount = count($parts); |
|
132
|
|
|
// check that the list count is an even number |
|
133
|
6 |
|
if (0 !== $partsCount % 2) { |
|
134
|
1 |
|
throw new SchemaValidatorException( |
|
135
|
1 |
|
"The schemaLocation value '" . $content . "' must have even number of URIs" |
|
136
|
|
|
); |
|
137
|
|
|
} |
|
138
|
|
|
// insert the uris pairs into the schemas |
|
139
|
5 |
|
for ($k = 0; $k < $partsCount; $k = $k + 2) { |
|
140
|
5 |
|
$schemas->create($parts[$k], $parts[$k + 1]); |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
5 |
|
return $schemas; |
|
145
|
|
|
} |
|
146
|
|
|
} |
|
147
|
|
|
|
eclipxe13 /
XmlSchemaValidator
If a method or function can return multiple different values and unless you are sure that you only can receive a single value in this context, we recommend to add an additional type check:
If this a common case that PHP Analyzer should handle natively, please let us know by opening an issue.