php写的设置过滤及保留属性的类

2. /**
3. @ 设置及保留属性的类
4. @ class cleanHtml
5. @ link：bbs.it-home.org
6. @ date：2013/2/28
7. */
8. function reg_escape( $str )
9. {
10. $conversions = array( "^" => "\^", "[" => "\[", "." => "\.", "$" => "\$", "{" => "\{", "*" => "\*", "(" => "\(", "\\" => "\\\\", "/" => "\/", "+" => "\+", ")" => "\)", "|" => "\|", "?" => "\?", "<" => "\<", ">" => "\>" );
11. return strtr( $str, $conversions );
12. }
14. /**
15. * Strip attribute Class
16. * Remove attributes from XML elements
17. * @author David (semlabs.co.uk)
18. * @version 0.2.1
19. */
21. class cleanHtml{
22. public $str = '';
23. public $allow = array();
24. public $exceptions = array();
25. public $ignore = array();
27. public function strip( $str )
28. {
29. $this->str = $str;
31. if( is_string( $str ) && strlen( $str ) > 0 )
32. {
33. $res = $this->findElements();
34. if( is_string( $res ) )
35. return $res;
36. $nodes = $this->findAttributes( $res );
37. $this->removeAttributes( $nodes );
38. }
40. return $this->str;
41. }
43. private function findElements()
44. {
46. # Create an array of elements with attributes
47. $nodes = array();
48. preg_match_all( "/<([^ !\/\>\n]+)([^>]*)>/i", $this->str, $elements );
49. foreach( $elements[1] as $el_key => $element )
50. {
51. if( $elements[2][$el_key] )
52. {
53. $literal = $elements[0][$el_key];
54. $element_name = $elements[1][$el_key];
55. $attributes = $elements[2][$el_key];
56. if( is_array( $this->ignore ) && !in_array( $element_name, $this->ignore ) )
57. $nodes[] = array( 'literal' => $literal, 'name' => $element_name, 'attributes' => $attributes );
58. }
59. }
61. # Return the XML if there were no attributes to remove
62. if( !$nodes[0] )
63. return $this->str;
64. else
65. return $nodes;
66. }
68. private function findAttributes( $nodes )
69. {
71. # Extract attributes
72. foreach( $nodes as &$node )
73. {
74. preg_match_all( "/([^ =]+)\s*=\s*[\"|']{0,1}([^\"']*)[\"|']{0,1}/i", $node['attributes'], $attributes );
75. if( $attributes[1] )
76. {
77. foreach( $attributes[1] as $att_key => $att )
78. {
79. $literal = $attributes[0][$att_key];
80. $attribute_name = $attributes[1][$att_key];
81. $value = $attributes[2][$att_key];
82. $atts[] = array( 'literal' => $literal, 'name' => $attribute_name, 'value' => $value );
83. }
84. }
85. else
86. $node['attributes'] = null;
88. $node['attributes'] = $atts;
89. unset( $atts );
90. }
92. return $nodes;
93. }
95. private function removeAttributes( $nodes )
96. {
98. # Remove unwanted attributes
99. foreach( $nodes as $node )
100. {
102. # Check if node has any attributes to be kept
103. $node_name = $node['name'];
104. $new_attributes = '';
105. if( is_array( $node['attributes'] ) )
106. {
107. foreach( $node['attributes'] as $attribute )
108. {
109. if( ( is_array( $this->allow ) && in_array( $attribute['name'], $this->allow ) ) || $this->isException( $node_name, $attribute['name'], $this->exceptions ) )
110. $new_attributes = $this->createAttributes( $new_attributes, $attribute['name'], $attribute['value'] );
111. }
112. }
113. $replacement = ( $new_attributes ) ? "<$node_name $new_attributes>" : "<$node_name>";
114. $this->str = preg_replace( '/'. reg_escape( $node['literal'] ) .'/', $replacement, $this->str );
115. }
116. }
118. private function isException( $element_name, $attribute_name, $exceptions )
119. {
120. if( array_key_exists($element_name, $this->exceptions) )
121. {
122. if( in_array( $attribute_name, $this->exceptions[$element_name] ) )
123. return true;
124. }
126. return false;
127. }
129. private function createAttributes( $new_attributes, $name, $value )
130. {
131. if( $new_attributes )
132. $new_attributes .= " ";
133. $new_attributes .= "$name=\"$value\"";
135. return $new_attributes;
136. }
137. }
138. ?>

调用实例：

2. $str = 'Here is some sample html that is getting broken ';
4. $sa = new cleanHtml;
5. $sa->allow = array( 'id' );
7. $sa->exceptions = array(
8. 'img' => array( 'src', 'alt' ),
9. 'a' => array( 'href', 'title' ),
10. 'iframe'=>array('src','frameborder'),
11. );
13. echo $str = $sa->strip( $str );
14. ?>