|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace Eclipxe\XmlResourceRetriever; |
|
6
|
|
|
|
|
7
|
|
|
use DOMDocument; |
|
8
|
|
|
use DOMElement; |
|
9
|
|
|
use finfo; |
|
10
|
|
|
use RuntimeException; |
|
11
|
|
|
|
|
12
|
|
|
/** |
|
13
|
|
|
* This is an abstract implementation of Retriever interface when working with XML contents |
|
14
|
|
|
* Both, XsdRetriever and XsltRetriver depends on this class |
|
15
|
|
|
*/ |
|
16
|
|
|
abstract class AbstractXmlRetriever extends AbstractBaseRetriever implements RetrieverInterface |
|
17
|
|
|
{ |
|
18
|
|
|
/** |
|
19
|
|
|
* Must return a string with the namespace to search for |
|
20
|
|
|
* |
|
21
|
|
|
* @return string |
|
22
|
|
|
*/ |
|
23
|
|
|
abstract protected function searchNamespace(): string; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Must return a table with rows (array of array) |
|
27
|
|
|
* every row must contain the keys element and attribute |
|
28
|
|
|
* "element" is the tag name to search for |
|
29
|
|
|
* "attribute" is the attribute name that contains the url |
|
30
|
|
|
* |
|
31
|
|
|
* @return array<array<string, string>> |
|
32
|
|
|
*/ |
|
33
|
|
|
abstract protected function searchElements(): array; |
|
34
|
|
|
|
|
35
|
7 |
|
public function retrieve(string $url): string |
|
36
|
|
|
{ |
|
37
|
7 |
|
$this->clearHistory(); |
|
38
|
7 |
|
return $this->doRetrieve($url); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @param string $resource |
|
43
|
|
|
* @return string |
|
44
|
|
|
*/ |
|
45
|
7 |
|
private function doRetrieve(string $resource): string |
|
46
|
|
|
{ |
|
47
|
7 |
|
$localFilename = $this->download($resource); |
|
48
|
7 |
|
$this->addToHistory($resource, $localFilename); |
|
49
|
|
|
|
|
50
|
7 |
|
$document = new DOMDocument(); |
|
51
|
|
|
// this error silenced call is intentional, |
|
52
|
|
|
// don't need to change the value of libxml_use_internal_errors for this |
|
53
|
|
|
/** @noinspection PhpUsageOfSilenceOperatorInspection */ |
|
54
|
7 |
|
$documentLoad = @$document->load($localFilename); |
|
55
|
7 |
|
if (false === $documentLoad) { |
|
56
|
2 |
|
unlink($localFilename); |
|
57
|
2 |
|
throw new RuntimeException("The source $resource contains errors"); |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
// call recursive get searching on specified the elements |
|
61
|
5 |
|
$changed = false; |
|
62
|
5 |
|
foreach ($this->searchElements() as $search) { |
|
63
|
5 |
|
$recursiveRetrieve = $this->recursiveRetrieve( |
|
64
|
5 |
|
$document, |
|
65
|
5 |
|
$search['element'], |
|
66
|
5 |
|
$search['attribute'], |
|
67
|
5 |
|
$resource, |
|
68
|
5 |
|
$localFilename |
|
69
|
5 |
|
); |
|
70
|
5 |
|
if ($recursiveRetrieve) { |
|
71
|
3 |
|
$changed = true; |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
5 |
|
if ($changed) { |
|
76
|
3 |
|
$document->save($localFilename); |
|
77
|
|
|
} |
|
78
|
5 |
|
return $localFilename; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
5 |
|
private function recursiveRetrieve( |
|
82
|
|
|
DOMDocument $document, |
|
83
|
|
|
string $tagName, |
|
84
|
|
|
string $attributeName, |
|
85
|
|
|
string $currentUrl, |
|
86
|
|
|
string $currentFile |
|
87
|
|
|
): bool { |
|
88
|
5 |
|
$modified = false; |
|
89
|
|
|
/** @var iterable<DOMElement> $elements */ |
|
90
|
5 |
|
$elements = $document->getElementsByTagNameNS($this->searchNamespace(), $tagName); |
|
91
|
5 |
|
foreach ($elements as $element) { |
|
92
|
3 |
|
if (! $element->hasAttribute($attributeName)) { |
|
93
|
1 |
|
continue; |
|
94
|
|
|
} |
|
95
|
3 |
|
$location = $element->getAttribute($attributeName); |
|
96
|
3 |
|
if ('' === $location) { |
|
97
|
1 |
|
continue; |
|
98
|
|
|
} |
|
99
|
3 |
|
$location = $this->relativeToAbsoluteUrl($location, $currentUrl); |
|
100
|
3 |
|
if (array_key_exists($location, $this->retrieveHistory())) { |
|
101
|
1 |
|
continue; |
|
102
|
|
|
} |
|
103
|
3 |
|
$downloadedChild = $this->doRetrieve($location); |
|
104
|
3 |
|
$relative = Utils::relativePath($currentFile, $downloadedChild); |
|
105
|
3 |
|
$element->setAttribute($attributeName, $relative); |
|
106
|
3 |
|
$modified = true; |
|
107
|
|
|
} |
|
108
|
5 |
|
return $modified; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* This method checks if the recently downloaded file from $source located at $path |
|
113
|
|
|
* is a valid resource, if not will remove the file and throw an exception |
|
114
|
|
|
* |
|
115
|
|
|
* @param string $source |
|
116
|
|
|
* @param string $localpath |
|
117
|
|
|
* @return void |
|
118
|
|
|
* @throws RuntimeException when the source is not valid |
|
119
|
|
|
*/ |
|
120
|
10 |
|
protected function checkIsValidDownloadedFile(string $source, string $localpath): void |
|
121
|
|
|
{ |
|
122
|
|
|
// check content is not empty |
|
123
|
10 |
|
if (0 === (int) filesize($localpath)) { |
|
124
|
1 |
|
unlink($localpath); |
|
125
|
1 |
|
throw new RuntimeException("The source $source is not an xml file because it is empty"); |
|
126
|
|
|
} |
|
127
|
|
|
// check content is xml |
|
128
|
9 |
|
$mimetype = strval((new finfo())->file($localpath, FILEINFO_MIME_TYPE)); |
|
129
|
9 |
|
if ('application/xml' !== $mimetype && 'text/' !== substr($mimetype, 0, 5)) { |
|
130
|
1 |
|
unlink($localpath); |
|
131
|
1 |
|
throw new RuntimeException("The source $source ($mimetype) is not an xml file"); |
|
132
|
|
|
} |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
3 |
|
private function relativeToAbsoluteUrl(string $url, string $currentUrl): string |
|
136
|
|
|
{ |
|
137
|
3 |
|
if (false !== $this->urlParts($url)) { |
|
138
|
3 |
|
return $url; |
|
139
|
|
|
} |
|
140
|
1 |
|
$currentParts = $this->urlParts($currentUrl); |
|
141
|
1 |
|
if (false === $currentParts) { |
|
142
|
|
|
$currentParts = []; |
|
143
|
|
|
} |
|
144
|
1 |
|
$currentParts['port'] = $currentParts['port'] ?? ''; |
|
145
|
1 |
|
$currentParts['port'] = ('' !== $currentParts['port']) ? ':' . $currentParts['port'] : ''; |
|
146
|
1 |
|
return implode('', [ |
|
147
|
1 |
|
$currentParts['scheme'], |
|
148
|
1 |
|
'://', |
|
149
|
1 |
|
$currentParts['host'], |
|
150
|
1 |
|
$currentParts['port'], |
|
151
|
1 |
|
implode('/', Utils::simplifyPath(dirname($currentParts['path']) . '/' . $url)), |
|
152
|
1 |
|
]); |
|
153
|
|
|
} |
|
154
|
|
|
} |
|
155
|
|
|
|
eclipxe13 /
XmlResourceRetriever
