1 |
85ad3d82
|
Assos Assos
|
<?php
|
2 |
|
|
|
3 |
|
|
/**
|
4 |
|
|
* @files
|
5 |
f066bdb5
|
Assos Assos
|
* Contains FeedsXPathParserXML.
|
6 |
|
|
*/
|
7 |
|
|
|
8 |
|
|
/**
|
9 |
|
|
* XPath parsing for XML.
|
10 |
85ad3d82
|
Assos Assos
|
*/
|
11 |
|
|
class FeedsXPathParserXML extends FeedsXPathParserBase {
|
12 |
|
|
|
13 |
|
|
/**
|
14 |
0b4524f6
|
Assos Assos
|
* {@inheritdoc}
|
15 |
85ad3d82
|
Assos Assos
|
*/
|
16 |
|
|
protected function setup($source_config, FeedsFetcherResult $fetcher_result) {
|
17 |
f066bdb5
|
Assos Assos
|
if (!empty($source_config['exp']['tidy']) && extension_loaded('tidy')) {
|
18 |
85ad3d82
|
Assos Assos
|
$config = array(
|
19 |
|
|
'input-xml' => TRUE,
|
20 |
|
|
'wrap' => 0,
|
21 |
|
|
'tidy-mark' => FALSE,
|
22 |
|
|
);
|
23 |
|
|
// Default tidy encoding is UTF8.
|
24 |
|
|
$encoding = $source_config['exp']['tidy_encoding'];
|
25 |
0b4524f6
|
Assos Assos
|
$raw = tidy_repair_string($fetcher_result->getRaw(), $config, $encoding);
|
26 |
85ad3d82
|
Assos Assos
|
}
|
27 |
|
|
else {
|
28 |
|
|
$raw = $fetcher_result->getRaw();
|
29 |
|
|
}
|
30 |
0b4524f6
|
Assos Assos
|
|
31 |
|
|
$options = LIBXML_NONET;
|
32 |
|
|
$options |= defined('LIBXML_COMPACT') ? LIBXML_COMPACT : 0;
|
33 |
|
|
$options |= defined('LIBXML_PARSEHUGE') ? LIBXML_PARSEHUGE : 0;
|
34 |
|
|
|
35 |
|
|
$document = new DOMDocument();
|
36 |
|
|
$document->strictErrorChecking = FALSE;
|
37 |
|
|
$document->recover = TRUE;
|
38 |
|
|
|
39 |
85ad3d82
|
Assos Assos
|
$use = $this->errorStart();
|
40 |
0b4524f6
|
Assos Assos
|
|
41 |
|
|
$success = $document->loadXML($raw, $options);
|
42 |
|
|
|
43 |
85ad3d82
|
Assos Assos
|
$this->errorStop($use, $source_config['exp']['errors']);
|
44 |
0b4524f6
|
Assos Assos
|
|
45 |
85ad3d82
|
Assos Assos
|
if (!$success) {
|
46 |
|
|
throw new Exception(t('There was an error parsing the XML document.'));
|
47 |
|
|
}
|
48 |
0b4524f6
|
Assos Assos
|
|
49 |
|
|
return $document;
|
50 |
85ad3d82
|
Assos Assos
|
}
|
51 |
|
|
|
52 |
0b4524f6
|
Assos Assos
|
/**
|
53 |
|
|
* {@inheritdoc}
|
54 |
|
|
*/
|
55 |
85ad3d82
|
Assos Assos
|
protected function getRaw(DOMNode $node) {
|
56 |
|
|
return $this->doc->saveXML($node);
|
57 |
|
|
}
|
58 |
0b4524f6
|
Assos Assos
|
|
59 |
85ad3d82
|
Assos Assos
|
} |