|
1
|
|
|
<?php |
|
2
|
|
|
namespace XmlResourceRetriever; |
|
3
|
|
|
|
|
4
|
|
|
use DOMDocument; |
|
5
|
|
|
use XmlResourceRetriever\Downloader\DownloaderInterface; |
|
6
|
|
|
use XmlResourceRetriever\Downloader\PhpDownloader; |
|
7
|
|
|
|
|
8
|
|
|
abstract class AbstractRetriever implements RetrieverInterface |
|
9
|
|
|
{ |
|
10
|
|
|
/** @var string */ |
|
11
|
|
|
private $basePath; |
|
12
|
|
|
|
|
13
|
|
|
/** @var DownloaderInterface */ |
|
14
|
|
|
private $downloader; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* This variable stores the list of retrieved resources to avoid infinite recursion |
|
18
|
|
|
* @var array |
|
19
|
|
|
*/ |
|
20
|
|
|
private $history = []; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* Must return a string with the namespace to search for |
|
24
|
|
|
* |
|
25
|
|
|
* @return string |
|
26
|
|
|
*/ |
|
27
|
|
|
abstract protected function searchNamespace(): string; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* Must return a table with rows (array of array) |
|
31
|
|
|
* every row must contain the keys element and attribute |
|
32
|
|
|
* "element" is the tag name to search for |
|
33
|
|
|
* "attribute" is the attribute name that contains the url |
|
34
|
|
|
* |
|
35
|
|
|
* @return array |
|
36
|
|
|
*/ |
|
37
|
|
|
abstract protected function searchElements(): array; |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* Retriever constructor. |
|
41
|
|
|
* |
|
42
|
|
|
* @param string $basePath |
|
43
|
|
|
* @param DownloaderInterface $downloader |
|
44
|
|
|
*/ |
|
45
|
16 |
|
public function __construct($basePath, DownloaderInterface $downloader = null) |
|
46
|
|
|
{ |
|
47
|
16 |
|
$this->basePath = $basePath; |
|
48
|
16 |
|
$this->setDownloader($downloader ? : new PhpDownloader()); |
|
49
|
16 |
|
} |
|
50
|
|
|
|
|
51
|
2 |
|
public function getBasePath(): string |
|
52
|
|
|
{ |
|
53
|
2 |
|
return $this->basePath; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
1 |
|
public function getDownloader(): DownloaderInterface |
|
57
|
|
|
{ |
|
58
|
1 |
|
return $this->downloader; |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
16 |
|
public function setDownloader(DownloaderInterface $downloader) |
|
62
|
|
|
{ |
|
63
|
16 |
|
$this->downloader = $downloader; |
|
64
|
16 |
|
} |
|
65
|
|
|
|
|
66
|
13 |
|
public function buildPath(string $url): string |
|
67
|
|
|
{ |
|
68
|
13 |
|
if (false === $parts = $this->urlParts($url)) { |
|
69
|
3 |
|
throw new \InvalidArgumentException("Invalid URL: $url"); |
|
70
|
|
|
} |
|
71
|
10 |
|
return $this->basePath . '/' . $parts['host'] . '/' . ltrim($parts['path'], '/'); |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
10 |
|
public function download(string $resource): string |
|
75
|
|
|
{ |
|
76
|
|
|
// validate resource |
|
77
|
10 |
|
if ('' === $resource) { |
|
78
|
1 |
|
throw new \UnexpectedValueException('The argument to download is empty'); |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
// set destination |
|
82
|
9 |
|
$localPath = $this->buildPath($resource); |
|
83
|
|
|
|
|
84
|
|
|
// create local path |
|
85
|
9 |
|
$dirname = dirname($localPath); |
|
86
|
9 |
|
if (! is_dir($dirname) && ! @mkdir($dirname, 0777, true)) { |
|
87
|
1 |
|
throw new \RuntimeException("Unable to create directory $dirname"); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
// download the file into its final destination |
|
91
|
8 |
|
$this->downloader->downloadTo($resource, $localPath); |
|
92
|
|
|
|
|
93
|
|
|
// check content is xml |
|
94
|
7 |
|
$mimetype = (new \finfo())->file($localPath, FILEINFO_MIME_TYPE); |
|
95
|
7 |
|
if (! in_array($mimetype, ['text/xml', 'application/xml'])) { |
|
96
|
1 |
|
unlink($localPath); |
|
97
|
1 |
|
throw new \RuntimeException("The source $resource ($mimetype) is not an xml file"); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
6 |
|
return $localPath; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
4 |
|
public function retrieve(string $resource): string |
|
104
|
|
|
{ |
|
105
|
4 |
|
$this->history = []; |
|
106
|
4 |
|
return $this->doRetrieve($resource); |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
1 |
|
public function retrieveHistory(): array |
|
110
|
|
|
{ |
|
111
|
1 |
|
return $this->history; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
/** |
|
115
|
|
|
* @param string $resource |
|
116
|
|
|
* @return string |
|
117
|
|
|
*/ |
|
118
|
4 |
|
private function doRetrieve(string $resource): string |
|
119
|
|
|
{ |
|
120
|
4 |
|
$localFilename = $this->download($resource); |
|
121
|
4 |
|
$this->history[$resource] = $localFilename; |
|
122
|
|
|
|
|
123
|
4 |
|
$document = new DOMDocument(); |
|
124
|
|
|
// this error silenced call is intentional, |
|
125
|
|
|
// don't need to change the value of libxml_use_internal_errors for this |
|
126
|
4 |
|
if (false === @$document->load($localFilename)) { |
|
127
|
1 |
|
unlink($localFilename); |
|
128
|
1 |
|
throw new \RuntimeException("The source $resource contains errors"); |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
// call recursive get searching on specified the elements |
|
132
|
3 |
|
$changed = false; |
|
133
|
3 |
|
foreach ($this->searchElements() as $search) { |
|
134
|
3 |
|
$recursiveRetrieve = $this->recursiveRetrieve( |
|
135
|
3 |
|
$document, |
|
136
|
3 |
|
$search['element'], |
|
137
|
3 |
|
$search['attribute'], |
|
138
|
3 |
|
$resource, |
|
139
|
3 |
|
$localFilename |
|
140
|
|
|
); |
|
141
|
3 |
|
if ($recursiveRetrieve) { |
|
142
|
3 |
|
$changed = true; |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
3 |
|
if ($changed) { |
|
147
|
3 |
|
$document->save($localFilename); |
|
148
|
|
|
} |
|
149
|
3 |
|
return $localFilename; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
3 |
|
private function recursiveRetrieve( |
|
153
|
|
|
DOMDocument $document, |
|
154
|
|
|
string $tagName, |
|
155
|
|
|
string $attributeName, |
|
156
|
|
|
string $currentUrl, |
|
157
|
|
|
string $currentFile |
|
158
|
|
|
): bool { |
|
159
|
3 |
|
$modified = false; |
|
160
|
3 |
|
$elements = $document->getElementsByTagNameNS($this->searchNamespace(), $tagName); |
|
161
|
3 |
|
foreach ($elements as $element) { |
|
162
|
|
|
/** @var \DOMElement $element */ |
|
163
|
3 |
|
if (! $element->hasAttribute($attributeName)) { |
|
164
|
1 |
|
continue; |
|
165
|
|
|
} |
|
166
|
3 |
|
$location = $element->getAttribute($attributeName); |
|
167
|
3 |
|
if ('' === $location) { |
|
168
|
1 |
|
continue; |
|
169
|
|
|
} |
|
170
|
3 |
|
$location = $this->relativeToAbsoluteUrl($location, $currentUrl); |
|
171
|
3 |
|
if (array_key_exists($location, $this->history)) { |
|
172
|
1 |
|
continue; |
|
173
|
|
|
} |
|
174
|
3 |
|
$downloadedChild = $this->doRetrieve($location); |
|
175
|
3 |
|
$relative = Utils::relativePath($currentFile, $downloadedChild); |
|
176
|
3 |
|
$element->setAttribute($attributeName, $relative); |
|
177
|
3 |
|
$modified = true; |
|
178
|
|
|
} |
|
179
|
3 |
|
return $modified; |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
13 |
|
private function urlParts(string $url) |
|
183
|
|
|
{ |
|
184
|
13 |
|
$options = FILTER_FLAG_SCHEME_REQUIRED | FILTER_FLAG_HOST_REQUIRED | FILTER_FLAG_PATH_REQUIRED; |
|
185
|
13 |
|
if (false === filter_var($url, FILTER_VALIDATE_URL, $options)) { |
|
186
|
4 |
|
return false; |
|
187
|
|
|
} |
|
188
|
10 |
|
return parse_url($url); |
|
189
|
|
|
} |
|
190
|
|
|
|
|
191
|
3 |
|
private function relativeToAbsoluteUrl(string $url, string $currentUrl) |
|
192
|
|
|
{ |
|
193
|
3 |
|
if (false !== $parts = $this->urlParts($url)) { |
|
194
|
3 |
|
return $url; |
|
195
|
|
|
} |
|
196
|
1 |
|
$currentParts = $this->urlParts($currentUrl); |
|
197
|
1 |
|
$currentParts['port'] = (isset($currentParts['port'])) ? ':' . $currentParts['port'] : ''; |
|
198
|
1 |
|
return implode('', [ |
|
199
|
1 |
|
$currentParts['scheme'], |
|
200
|
1 |
|
'://', |
|
201
|
1 |
|
$currentParts['host'], |
|
202
|
1 |
|
$currentParts['port'], |
|
203
|
1 |
|
implode('/', Utils::simplifyPath(dirname($currentParts['path']) . '/' . $url)), |
|
204
|
|
|
]); |
|
205
|
|
|
} |
|
206
|
|
|
} |
|
207
|
|
|
|
eclipxe13 /
XmlResourceRetriever
