|
1
|
|
|
<?php |
|
2
|
|
|
namespace Wa72\HtmlPageDom; |
|
3
|
|
|
|
|
4
|
|
|
/** |
|
5
|
|
|
* Static helper functions for HtmlPageDom |
|
6
|
|
|
* |
|
7
|
|
|
* @package Wa72\HtmlPageDom |
|
8
|
|
|
*/ |
|
9
|
|
|
class Helpers { |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* remove newlines from string and minimize whitespace (multiple whitespace characters replaced by one space) |
|
13
|
|
|
* useful for cleaning up text retrieved by HtmlPageCrawler::text() (nodeValue of a DOMNode) |
|
14
|
|
|
* |
|
15
|
|
|
* @param string $string |
|
16
|
|
|
* @return string |
|
17
|
|
|
*/ |
|
18
|
1 |
|
public static function trimNewlines($string) |
|
19
|
|
|
{ |
|
20
|
1 |
|
$string = str_replace("\n", ' ', $string); |
|
21
|
1 |
|
$string = str_replace("\r", ' ', $string); |
|
22
|
1 |
|
$string = preg_replace('/\s+/', ' ', $string); |
|
23
|
1 |
|
return trim($string); |
|
24
|
|
|
} |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Convert CSS string to array |
|
28
|
|
|
* |
|
29
|
|
|
* @param string $css list of CSS properties separated by ; |
|
30
|
|
|
* @return array name=>value pairs of CSS properties |
|
31
|
|
|
*/ |
|
32
|
1 |
|
public static function cssStringToArray($css) |
|
33
|
|
|
{ |
|
34
|
1 |
|
$statements = explode(';', preg_replace('/\s+/s', ' ', $css)); |
|
35
|
1 |
|
$styles = array(); |
|
36
|
1 |
|
foreach ($statements as $statement) { |
|
37
|
1 |
|
$statement = trim($statement); |
|
38
|
1 |
|
if ('' === $statement) { |
|
39
|
1 |
|
continue; |
|
40
|
|
|
} |
|
41
|
1 |
|
$p = strpos($statement, ':'); |
|
42
|
1 |
|
if ($p <= 0) { |
|
43
|
1 |
|
continue; |
|
44
|
|
|
} // invalid statement, just ignore it |
|
45
|
1 |
|
$key = trim(substr($statement, 0, $p)); |
|
46
|
1 |
|
$value = trim(substr($statement, $p + 1)); |
|
47
|
1 |
|
$styles[$key] = $value; |
|
48
|
1 |
|
} |
|
49
|
1 |
|
return $styles; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* Convert CSS name->value array to string |
|
54
|
|
|
* |
|
55
|
|
|
* @param array $array name=>value pairs of CSS properties |
|
56
|
|
|
* @return string list of CSS properties separated by ; |
|
57
|
|
|
*/ |
|
58
|
1 |
|
public static function cssArrayToString($array) |
|
59
|
|
|
{ |
|
60
|
1 |
|
$styles = ''; |
|
61
|
1 |
|
foreach ($array as $key => $value) { |
|
62
|
1 |
|
$styles .= $key . ': ' . $value . ';'; |
|
63
|
1 |
|
} |
|
64
|
1 |
|
return $styles; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* Helper function for getting a body element |
|
69
|
|
|
* from an HTML fragment |
|
70
|
|
|
* |
|
71
|
|
|
* @param string $html A fragment of HTML code |
|
72
|
|
|
* @param string $charset |
|
73
|
|
|
* @return \DOMNode The body node containing child nodes created from the HTML fragment |
|
74
|
|
|
*/ |
|
75
|
15 |
|
public static function getBodyNodeFromHtmlFragment($html, $charset = 'UTF-8') |
|
76
|
|
|
{ |
|
77
|
15 |
|
$html = '<html><body>' . $html . '</body></html>'; |
|
78
|
15 |
|
$current = libxml_use_internal_errors(true); |
|
79
|
15 |
|
$disableEntities = libxml_disable_entity_loader(true); |
|
80
|
15 |
|
$d = new \DOMDocument('1.0', $charset); |
|
81
|
15 |
|
$d->validateOnParse = true; |
|
82
|
15 |
|
if (function_exists('mb_convert_encoding') && in_array( |
|
83
|
15 |
|
strtolower($charset), |
|
84
|
15 |
|
array_map('strtolower', mb_list_encodings()) |
|
85
|
15 |
|
) |
|
86
|
15 |
|
) { |
|
87
|
15 |
|
$html = mb_convert_encoding($html, 'HTML-ENTITIES', $charset); |
|
88
|
15 |
|
} |
|
89
|
15 |
|
@$d->loadHTML($html); |
|
|
|
|
|
|
90
|
15 |
|
libxml_use_internal_errors($current); |
|
91
|
15 |
|
libxml_disable_entity_loader($disableEntities); |
|
92
|
15 |
|
return $d->getElementsByTagName('body')->item(0); |
|
93
|
|
|
} |
|
94
|
|
|
} |
|
95
|
|
|
|
wasinger /
htmlpagedom
If you suppress an error, we recommend checking for the error condition explicitly: