|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* RemoveComments |
|
4
|
|
|
* |
|
5
|
|
|
* @author Tim Lochmüller |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace HTML\Sourceopt\Manipulation; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* RemoveComments |
|
12
|
|
|
*/ |
|
13
|
|
|
class RemoveComments implements ManipulationInterface |
|
14
|
|
|
{ |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Patterns for white-listing comments inside content |
|
18
|
|
|
* |
|
19
|
|
|
* @var array |
|
20
|
|
|
*/ |
|
21
|
|
|
protected $whiteListCommentsPatterns = []; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @param string $html The original HTML |
|
25
|
|
|
* @param array $configuration Configuration |
|
26
|
|
|
* |
|
27
|
|
|
* @return string the manipulated HTML |
|
28
|
|
|
*/ |
|
29
|
|
|
public function manipulate($html, array $configuration = []) |
|
30
|
|
|
{ |
|
31
|
|
|
if (isset($configuration['keep.'])) { |
|
32
|
|
|
$this->whiteListCommentsPatterns = $configuration['keep.']; |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
// match all styles, scripts and comments |
|
36
|
|
|
$matches = []; |
|
37
|
|
|
preg_match_all( |
|
38
|
|
|
'/(?s)((<!--.*?-->)|(<[ \n\r]*style[^>]*>.*?<[ \n\r]*\/style[^>]*>)|(<[ \n\r]*script[^>]*>.*?<[ \n\r]*\/script[^>]*>))/im', |
|
39
|
|
|
$html, |
|
40
|
|
|
$matches |
|
41
|
|
|
); |
|
42
|
|
|
foreach ($matches[0] as $tag) { |
|
43
|
|
|
if ($this->keepComment($tag) === false) { |
|
44
|
|
|
$html = str_replace($tag, '', $html); |
|
45
|
|
|
} |
|
46
|
|
|
} |
|
47
|
|
|
return $html; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Check if a comment is defined to be kept in a pattern whiteListOfComments |
|
52
|
|
|
* |
|
53
|
|
|
* @param string $commentHtml |
|
54
|
|
|
* |
|
55
|
|
|
* @return boolean |
|
56
|
|
|
*/ |
|
57
|
|
|
protected function keepComment($commentHtml) |
|
58
|
|
|
{ |
|
59
|
|
|
// if not even a comment, skip this |
|
60
|
|
|
if (!preg_match('/^\<\!\-\-(.*?)\-\-\>$/usi', $commentHtml)) { |
|
61
|
|
|
return true; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
// if not defined in white list |
|
65
|
|
|
if (!empty($this->whiteListCommentsPatterns)) { |
|
66
|
|
|
$commentHtml = str_replace("<!--", "", $commentHtml); |
|
67
|
|
|
$commentHtml = str_replace("-->", "", $commentHtml); |
|
68
|
|
|
$commentHtml = trim($commentHtml); |
|
69
|
|
|
foreach ($this->whiteListCommentsPatterns as $pattern) { |
|
70
|
|
|
if (preg_match($pattern, $commentHtml)) { |
|
71
|
|
|
return true; |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
return false; |
|
76
|
|
|
} |
|
77
|
|
|
} |
|
78
|
|
|
|