|
1
|
|
|
<?php |
|
2
|
|
|
namespace EZAMA{ |
|
3
|
|
|
class BeforeStrip |
|
4
|
|
|
{ |
|
5
|
|
|
protected $multi=false; |
|
6
|
|
|
protected $is_html=false; |
|
7
|
|
|
protected $is_php=false; |
|
8
|
|
|
protected $html=array(); |
|
9
|
|
|
protected $php=array(); |
|
10
|
|
|
protected $content=''; |
|
11
|
|
|
protected $doctype; |
|
12
|
|
|
protected $body; |
|
13
|
|
|
protected $head; |
|
14
|
|
|
protected $html_tag; |
|
15
|
|
|
public function __construct($html) |
|
16
|
|
|
{ |
|
17
|
|
|
if (!is_string($html)) { |
|
18
|
|
|
throw new \InvalidArgumentException(sprintf('Waiting string, %s given', gettype($html))); |
|
19
|
|
|
} |
|
20
|
|
|
$html=is_file($html)?file_get_contents($html):$html; |
|
21
|
|
|
$this->setMainTags($html); |
|
22
|
|
|
|
|
23
|
|
|
$html='<div>'.str_replace( |
|
24
|
|
|
array($this->doctype,$this->html_tag,'</html>',$this->head,'</head>',$this->body,'</body>'), |
|
25
|
|
|
array('<doctypetag '.substr($this->doctype, 10),'<htmltag '.substr($this->html_tag, 5),'</htmltag>','<headtag '.substr($this->head, 5),'</headtag>','<bodytag '.substr($this->body, 5),'</bodytag>'), |
|
26
|
|
|
$html |
|
27
|
|
|
).'</doctypetag></div>'; |
|
28
|
|
|
$preprocessed=token_get_all($html); |
|
29
|
|
|
|
|
30
|
|
|
$HTML=array_filter($preprocessed, function ($v) { |
|
31
|
|
|
return is_array($v)&&$v[0]===T_INLINE_HTML; |
|
32
|
|
|
}); |
|
33
|
|
|
$PHP=array_diff_key($preprocessed, $HTML); |
|
34
|
|
|
$this->init($HTML, $PHP, $html); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
protected function setMainTags(&$html) |
|
38
|
|
|
{ |
|
39
|
|
|
$doctypeOffset=stripos($html, '<!doctype '); |
|
40
|
|
|
$headOffset=stripos($html, '<head'); |
|
41
|
|
|
$htmlTagOffset=stripos($html, '<html'); |
|
42
|
|
|
$bodyOffset=stripos($html, '<body'); |
|
43
|
|
|
if (false!==$doctypeOffset) { |
|
44
|
|
|
$endDoctypeOffset=strpos($html, '>', $doctypeOffset); |
|
45
|
|
|
$this->doctype=substr($html, $doctypeOffset, $endDoctypeOffset-$doctypeOffset+1); |
|
46
|
|
|
} |
|
47
|
|
|
if (false!==$headOffset) { |
|
48
|
|
|
$endHeadOffset=strpos($html, '>', $headOffset); |
|
49
|
|
|
$this->head=substr($html, $headOffset, $endHeadOffset-$headOffset+1); |
|
50
|
|
|
} |
|
51
|
|
|
if (false!==$bodyOffset) { |
|
52
|
|
|
$endBodyOffset=strpos($html, '>', $bodyOffset); |
|
53
|
|
|
$this->body=substr($html, $bodyOffset, $endBodyOffset-$bodyOffset+1); |
|
54
|
|
|
} |
|
55
|
|
|
if (false!==$htmlTagOffset) { |
|
56
|
|
|
$endHtmlTagOffset=strpos($html, '>', $htmlTagOffset); |
|
57
|
|
|
$this->html_tag=substr($html, $htmlTagOffset, $endHtmlTagOffset-$htmlTagOffset+1); |
|
58
|
|
|
} |
|
59
|
|
|
} |
|
60
|
|
|
protected function init(&$HTML, &$PHP, &$html) |
|
61
|
|
|
{ |
|
62
|
|
|
list($is_h, $is_p)=array((bool)$HTML,(bool)$PHP); |
|
63
|
|
|
list($this->is_html, $this->is_php, $this->multi, $this->content, $this->html, $this->php)=array($is_h,$is_p,$is_h&&$is_p,$html,$is_p?$HTML:array(),$is_p?$PHP:array()); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
public function mustLoadMulti() |
|
67
|
|
|
{ |
|
68
|
|
|
return $this->multi; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
public function isHtml() |
|
72
|
|
|
{ |
|
73
|
|
|
return $this->is_html; |
|
74
|
|
|
} |
|
75
|
|
|
public function isPHP() |
|
76
|
|
|
{ |
|
77
|
|
|
return $this->is_php; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
public function getPHP() |
|
81
|
|
|
{ |
|
82
|
|
|
return $this->php; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
public function getHTML() |
|
86
|
|
|
{ |
|
87
|
|
|
return $this->html; |
|
88
|
|
|
} |
|
89
|
|
|
public function getContent() |
|
90
|
|
|
{ |
|
91
|
|
|
return $this->content; |
|
92
|
|
|
} |
|
93
|
|
|
public function getDoctype() |
|
94
|
|
|
{ |
|
95
|
|
|
return $this->doctype; |
|
96
|
|
|
} |
|
97
|
|
|
public function getHead() |
|
98
|
|
|
{ |
|
99
|
|
|
return $this->head; |
|
100
|
|
|
} |
|
101
|
|
|
public function getBody() |
|
102
|
|
|
{ |
|
103
|
|
|
return $this->body; |
|
104
|
|
|
} |
|
105
|
|
|
public function getHtmlTag() |
|
106
|
|
|
{ |
|
107
|
|
|
return $this->html_tag; |
|
108
|
|
|
} |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
manuwhat /
strip-tags
