|
1
|
<?php
|
|
2
|
|
|
3
|
/**
|
|
4
|
* @files
|
|
5
|
* Contains FeedsXPathParserHTML.
|
|
6
|
*/
|
|
7
|
|
|
8
|
/**
|
|
9
|
* XPath parsing for HTML.
|
|
10
|
*/
|
|
11
|
class FeedsXPathParserHTML extends FeedsXPathParserBase {
|
|
12
|
|
|
13
|
/**
|
|
14
|
* Whether this version of PHP has a useable saveHTML() method.
|
|
15
|
*
|
|
16
|
* @var bool
|
|
17
|
*/
|
|
18
|
protected $hasSaveHTML = FALSE;
|
|
19
|
|
|
20
|
/**
|
|
21
|
* {@inheritdoc}
|
|
22
|
*/
|
|
23
|
public function __construct($id) {
|
|
24
|
parent::__construct($id);
|
|
25
|
|
|
26
|
// DOMDocument::saveHTML() cannot take $node as an argument prior to 5.3.6.
|
|
27
|
if (version_compare(phpversion(), '5.3.6', '>=')) {
|
|
28
|
$this->hasSaveHTML = TRUE;
|
|
29
|
}
|
|
30
|
}
|
|
31
|
|
|
32
|
/**
|
|
33
|
* {@inheritdoc}
|
|
34
|
*/
|
|
35
|
protected function setup($source_config, FeedsFetcherResult $fetcher_result) {
|
|
36
|
if (!empty($source_config['exp']['tidy']) && extension_loaded('tidy')) {
|
|
37
|
$config = array(
|
|
38
|
'merge-divs' => FALSE,
|
|
39
|
'merge-spans' => FALSE,
|
|
40
|
'join-styles' => FALSE,
|
|
41
|
'drop-empty-paras' => FALSE,
|
|
42
|
'wrap' => 0,
|
|
43
|
'tidy-mark' => FALSE,
|
|
44
|
'escape-cdata' => TRUE,
|
|
45
|
'word-2000' => TRUE,
|
|
46
|
);
|
|
47
|
// Default tidy encoding is UTF8.
|
|
48
|
$encoding = $source_config['exp']['tidy_encoding'];
|
|
49
|
$raw = tidy_repair_string($fetcher_result->getRaw(), $config, $encoding);
|
|
50
|
}
|
|
51
|
else {
|
|
52
|
$raw = $fetcher_result->getRaw();
|
|
53
|
}
|
|
54
|
|
|
55
|
$document = new DOMDocument();
|
|
56
|
$document->strictErrorChecking = FALSE;
|
|
57
|
$document->recover = TRUE;
|
|
58
|
|
|
59
|
// Use our own error handling.
|
|
60
|
$use = $this->errorStart();
|
|
61
|
|
|
62
|
if (version_compare(PHP_VERSION, '5.4.0', '>=')) {
|
|
63
|
$options = LIBXML_NONET;
|
|
64
|
$options |= defined('LIBXML_COMPACT') ? LIBXML_COMPACT : 0;
|
|
65
|
$options |= defined('LIBXML_PARSEHUGE') ? LIBXML_PARSEHUGE : 0;
|
|
66
|
|
|
67
|
$success = $document->loadHTML($raw, $options);
|
|
68
|
}
|
|
69
|
else {
|
|
70
|
$success = $document->loadHTML($raw);
|
|
71
|
}
|
|
72
|
|
|
73
|
$this->errorStop($use, $source_config['exp']['errors']);
|
|
74
|
|
|
75
|
if (!$success) {
|
|
76
|
throw new Exception(t('There was an error parsing the HTML document.'));
|
|
77
|
}
|
|
78
|
|
|
79
|
return $document;
|
|
80
|
}
|
|
81
|
|
|
82
|
/**
|
|
83
|
* {@inheritdoc}
|
|
84
|
*/
|
|
85
|
protected function getRaw(DOMNode $node) {
|
|
86
|
if ($this->hasSaveHTML) {
|
|
87
|
return $this->doc->saveHTML($node);
|
|
88
|
}
|
|
89
|
|
|
90
|
return $this->doc->saveXML($node, LIBXML_NOEMPTYTAG);
|
|
91
|
}
|
|
92
|
|
|
93
|
}
|
