|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace simplehtmldom; |
|
4
|
|
|
|
|
5
|
|
|
/* |
|
6
|
|
|
* Website: http://sourceforge.net/projects/simplehtmldom/ |
|
7
|
|
|
* Acknowledge: Jose Solorzano (https://sourceforge.net/projects/php-html/) |
|
8
|
|
|
* |
|
9
|
|
|
* Licensed under The MIT License |
|
10
|
|
|
* See the LICENSE file in the project root for more information. |
|
11
|
|
|
* |
|
12
|
|
|
* Authors: |
|
13
|
|
|
* S.C. Chen |
|
14
|
|
|
* John Schlick |
|
15
|
|
|
* Rus Carroll |
|
16
|
|
|
* logmanoriginal |
|
17
|
|
|
* |
|
18
|
|
|
* Contributors: |
|
19
|
|
|
* Yousuke Kumakura |
|
20
|
|
|
* Vadim Voituk |
|
21
|
|
|
* Antcs |
|
22
|
|
|
* |
|
23
|
|
|
* Version Rev. 2.0-RC2 (415) |
|
24
|
|
|
*/ |
|
25
|
|
|
|
|
26
|
|
|
//require_once __DIR__ . '/constants.php'; |
|
27
|
|
|
include_once __DIR__ . '/HtmlDocument.php'; |
|
28
|
|
|
include_once __DIR__ . '/HtmlNode.php'; |
|
29
|
|
|
|
|
30
|
|
|
if (!defined('DEFAULT_TARGET_CHARSET') && defined('\simplehtmldom\DEFAULT_TARGET_CHARSET')) { |
|
31
|
|
|
define('DEFAULT_TARGET_CHARSET', \simplehtmldom\DEFAULT_TARGET_CHARSET); |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
if (!defined('DEFAULT_BR_TEXT') && defined('\simplehtmldom\DEFAULT_BR_TEXT')) { |
|
35
|
|
|
define('DEFAULT_BR_TEXT', \simplehtmldom\DEFAULT_BR_TEXT); |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
if (!defined('DEFAULT_SPAN_TEXT') && defined('\simplehtmldom\DEFAULT_SPAN_TEXT')) { |
|
39
|
|
|
define('DEFAULT_SPAN_TEXT', \simplehtmldom\DEFAULT_SPAN_TEXT); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
if (!defined('MAX_FILE_SIZE') && defined('\simplehtmldom\MAX_FILE_SIZE')) { |
|
43
|
|
|
define('MAX_FILE_SIZE', \simplehtmldom\MAX_FILE_SIZE); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
if (!defined('DEFAULT_TARGET_CHARSET')) { |
|
47
|
|
|
define('DEFAULT_TARGET_CHARSET', 'UTF-8'); |
|
48
|
|
|
} |
|
49
|
|
|
if (!defined('DEFAULT_BR_TEXT')) { |
|
50
|
|
|
define('DEFAULT_BR_TEXT', "\r\n"); |
|
51
|
|
|
} |
|
52
|
|
|
if (!defined('DEFAULT_SPAN_TEXT')) { |
|
53
|
|
|
define('DEFAULT_SPAN_TEXT', ' '); |
|
54
|
|
|
} |
|
55
|
|
|
if (!defined('MAX_FILE_SIZE')) { |
|
56
|
|
|
define('MAX_FILE_SIZE', 2621440); |
|
57
|
|
|
} |
|
58
|
|
|
if (!defined('HDOM_SMARTY_AS_TEXT')) { |
|
59
|
|
|
define('HDOM_SMARTY_AS_TEXT', 1); |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
define('HDOM_TYPE_ELEMENT', \simplehtmldom\HtmlNode::HDOM_TYPE_ELEMENT); |
|
63
|
|
|
define('HDOM_TYPE_COMMENT', \simplehtmldom\HtmlNode::HDOM_TYPE_COMMENT); |
|
64
|
|
|
define('HDOM_TYPE_TEXT', \simplehtmldom\HtmlNode::HDOM_TYPE_TEXT); |
|
65
|
|
|
define('HDOM_TYPE_ROOT', \simplehtmldom\HtmlNode::HDOM_TYPE_ROOT); |
|
66
|
|
|
define('HDOM_TYPE_UNKNOWN', \simplehtmldom\HtmlNode::HDOM_TYPE_UNKNOWN); |
|
67
|
|
|
define('HDOM_QUOTE_DOUBLE', \simplehtmldom\HtmlNode::HDOM_QUOTE_DOUBLE); |
|
68
|
|
|
define('HDOM_QUOTE_SINGLE', \simplehtmldom\HtmlNode::HDOM_QUOTE_SINGLE); |
|
69
|
|
|
define('HDOM_QUOTE_NO', \simplehtmldom\HtmlNode::HDOM_QUOTE_NO); |
|
70
|
|
|
define('HDOM_INFO_BEGIN', \simplehtmldom\HtmlNode::HDOM_INFO_BEGIN); |
|
71
|
|
|
define('HDOM_INFO_END', \simplehtmldom\HtmlNode::HDOM_INFO_END); |
|
72
|
|
|
define('HDOM_INFO_QUOTE', \simplehtmldom\HtmlNode::HDOM_INFO_QUOTE); |
|
73
|
|
|
define('HDOM_INFO_SPACE', \simplehtmldom\HtmlNode::HDOM_INFO_SPACE); |
|
74
|
|
|
define('HDOM_INFO_TEXT', \simplehtmldom\HtmlNode::HDOM_INFO_TEXT); |
|
75
|
|
|
define('HDOM_INFO_INNER', \simplehtmldom\HtmlNode::HDOM_INFO_INNER); |
|
76
|
|
|
define('HDOM_INFO_OUTER', \simplehtmldom\HtmlNode::HDOM_INFO_OUTER); |
|
77
|
|
|
define('HDOM_INFO_ENDSPACE', \simplehtmldom\HtmlNode::HDOM_INFO_ENDSPACE); |
|
78
|
|
|
//define('HDOM_SMARTY_AS_TEXT', HDOM_SMARTY_AS_TEXT); |
|
79
|
|
|
class_alias('\simplehtmldom\HtmlDocument', 'simple_html_dom', true); |
|
80
|
|
|
//class_alias('\simplehtmldom\HtmlNode', 'simple_html_dom_node', true); |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* String to get html dom. |
|
84
|
|
|
* |
|
85
|
|
|
* @param string $str HTML string |
|
86
|
|
|
*/ |
|
87
|
|
|
class simple_html_dom // extends HtmlDocument |
|
88
|
|
|
{ |
|
89
|
|
|
public $innertext; |
|
90
|
|
|
public $outertext; |
|
91
|
|
|
|
|
92
|
|
|
public function __construct() |
|
93
|
|
|
{ |
|
94
|
|
|
//parent::__construct(); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
public function file_get_html( |
|
98
|
|
|
$url, |
|
99
|
|
|
$use_include_path = false, |
|
100
|
|
|
$context = null, |
|
101
|
|
|
$offset = 0, |
|
102
|
|
|
$maxLen = -1, |
|
103
|
|
|
$lowercase = true, |
|
104
|
|
|
$forceTagsClosed = true, |
|
105
|
|
|
$target_charset = \simplehtmldom\DEFAULT_TARGET_CHARSET, |
|
106
|
|
|
$stripRN = true, |
|
107
|
|
|
$defaultBRText = \simplehtmldom\DEFAULT_BR_TEXT, |
|
108
|
|
|
$defaultSpanText = \simplehtmldom\DEFAULT_SPAN_TEXT |
|
109
|
|
|
) { |
|
110
|
|
|
if ($maxLen <= 0) { |
|
111
|
|
|
$maxLen = \simplehtmldom\MAX_FILE_SIZE; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
$dom = new HtmlDocument( |
|
115
|
|
|
null, |
|
116
|
|
|
$lowercase, |
|
117
|
|
|
$forceTagsClosed, |
|
118
|
|
|
$target_charset, |
|
119
|
|
|
$stripRN, |
|
120
|
|
|
$defaultBRText, |
|
121
|
|
|
$defaultSpanText |
|
122
|
|
|
); |
|
123
|
|
|
|
|
124
|
|
|
$contents = file_get_contents( |
|
125
|
|
|
$url, |
|
126
|
|
|
$use_include_path, |
|
127
|
|
|
$context, |
|
128
|
|
|
$offset, |
|
129
|
|
|
$maxLen + 1 // Load extra byte for limit check |
|
130
|
|
|
); |
|
131
|
|
|
|
|
132
|
|
|
if (empty($contents) || strlen($contents) > $maxLen) { |
|
133
|
|
|
$dom->clear(); |
|
|
|
|
|
|
134
|
|
|
|
|
135
|
|
|
return false; |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
return $dom->load($contents, $lowercase, $stripRN); |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
public function str_get_html( |
|
142
|
|
|
$str, |
|
143
|
|
|
$lowercase = true, |
|
144
|
|
|
$forceTagsClosed = true, |
|
145
|
|
|
$target_charset = DEFAULT_TARGET_CHARSET, |
|
146
|
|
|
$stripRN = true, |
|
147
|
|
|
$defaultBRText = DEFAULT_BR_TEXT, |
|
148
|
|
|
$defaultSpanText = DEFAULT_SPAN_TEXT |
|
149
|
|
|
) { |
|
150
|
|
|
$dom = new HtmlDocument( |
|
151
|
|
|
null, |
|
152
|
|
|
$lowercase, |
|
153
|
|
|
$forceTagsClosed, |
|
154
|
|
|
$target_charset, |
|
155
|
|
|
$stripRN, |
|
156
|
|
|
$defaultBRText, |
|
157
|
|
|
$defaultSpanText |
|
158
|
|
|
); |
|
159
|
|
|
|
|
160
|
|
|
if (empty($str) || strlen($str) > MAX_FILE_SIZE) { |
|
161
|
|
|
$dom->clear(); |
|
162
|
|
|
|
|
163
|
|
|
return false; |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
return $dom->load($str, $lowercase, $stripRN); |
|
167
|
|
|
} |
|
168
|
|
|
} |
|
169
|
|
|
|
|
170
|
|
|
/** @codeCoverageIgnore */ |
|
171
|
|
|
function dump_html_tree($node, $show_attr = true, $deep = 0) |
|
|
|
|
|
|
172
|
|
|
{ |
|
173
|
|
|
$node->dump($node); |
|
174
|
|
|
} |
|
175
|
|
|
|