lib/htmlpurifier/HTMLPurifier/HTMLModule/TransformToStrict.php

   1 <?php
   2
   3 require_once 'HTMLPurifier/ChildDef/StrictBlockquote.php';
   4
   5 require_once 'HTMLPurifier/TagTransform/Simple.php';
   6 require_once 'HTMLPurifier/TagTransform/Center.php';
   7 require_once 'HTMLPurifier/TagTransform/Font.php';
   8
   9 require_once 'HTMLPurifier/AttrTransform/Lang.php';
  10 require_once 'HTMLPurifier/AttrTransform/BgColor.php';
  11 require_once 'HTMLPurifier/AttrTransform/BoolToCSS.php';
  12 require_once 'HTMLPurifier/AttrTransform/Border.php';
  13 require_once 'HTMLPurifier/AttrTransform/Name.php';
  14 require_once 'HTMLPurifier/AttrTransform/Length.php';
  15 require_once 'HTMLPurifier/AttrTransform/ImgSpace.php';
  16 require_once 'HTMLPurifier/AttrTransform/EnumToCSS.php';
  17
  18 /**
  19  * Proprietary module that transforms deprecated elements into Strict
  20  * HTML (see HTML 4.01 and XHTML 1.0) when possible.
  21  */
  22
  23 class HTMLPurifier_HTMLModule_TransformToStrict extends HTMLPurifier_HTMLModule
  24 {
  25
  26     var $name = 'TransformToStrict';
  27
  28     // we're actually modifying these elements, not defining them
  29     var $elements = array('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p',
  30         'blockquote', 'table', 'td', 'th', 'tr', 'img', 'a', 'hr', 'br',
  31         'caption', 'ul', 'ol', 'li');
  32
  33     var $info_tag_transform = array(
  34         // placeholders, see constructor for definitions
  35         'font'  => false,
  36         'menu'  => false,
  37         'dir'   => false,
  38         'center'=> false
  39     );
  40
  41     var $attr_collections = array(
  42         'Lang' => array(
  43             'lang' => false // placeholder
  44         )
  45     );
  46
  47     var $info_attr_transform_post = array(
  48         'lang' => false // placeholder
  49     );
  50
  51     function HTMLPurifier_HTMLModule_TransformToStrict() {
  52
  53         // behavior with transformations when there's another CSS property
  54         // working on it is interesting: the CSS will *always* override
  55         // the deprecated attribute, whereas an inline CSS declaration will
  56         // override the corresponding declaration in, say, an external
  57         // stylesheet. This behavior won't affect most people, but it
  58         // does represent an operational difference we CANNOT fix.
  59
  60         // deprecated tag transforms
  61         $this->info_tag_transform['font']   = new HTMLPurifier_TagTransform_Font();
  62         $this->info_tag_transform['menu']   = new HTMLPurifier_TagTransform_Simple('ul');
  63         $this->info_tag_transform['dir']    = new HTMLPurifier_TagTransform_Simple('ul');
  64         $this->info_tag_transform['center'] = new HTMLPurifier_TagTransform_Center();
  65
  66         foreach ($this->elements as $name) {
  67             $this->info[$name] = new HTMLPurifier_ElementDef();
  68             $this->info[$name]->standalone = false;
  69         }
  70
  71         // deprecated attribute transforms
  72
  73         // align battery
  74         $align_lookup = array();
  75         $align_values = array('left', 'right', 'center', 'justify');
  76         foreach ($align_values as $v) $align_lookup[$v] = "text-align:$v;";
  77         $this->info['h1']->attr_transform_pre['align'] =
  78         $this->info['h2']->attr_transform_pre['align'] =
  79         $this->info['h3']->attr_transform_pre['align'] =
  80         $this->info['h4']->attr_transform_pre['align'] =
  81         $this->info['h5']->attr_transform_pre['align'] =
  82         $this->info['h6']->attr_transform_pre['align'] =
  83         $this->info['p'] ->attr_transform_pre['align'] =
  84                     new HTMLPurifier_AttrTransform_EnumToCSS('align', $align_lookup);
  85
  86         // xml:lang <=> lang mirroring, implement in TransformToStrict,
  87         // this is overridden in TransformToXHTML11
  88         $this->info_attr_transform_post['lang'] = new HTMLPurifier_AttrTransform_Lang();
  89         $this->attr_collections['Lang']['lang'] = new HTMLPurifier_AttrDef_Lang();
  90
  91         // this should not be applied to XHTML 1.0 Transitional, ONLY
  92         // XHTML 1.0 Strict. We may need three classes
  93         $this->info['blockquote']->content_model_type = 'strictblockquote';
  94         $this->info['blockquote']->child = false; // recalculate please!
  95
  96         $this->info['table']->attr_transform_pre['bgcolor'] =
  97         $this->info['tr']->attr_transform_pre['bgcolor'] =
  98         $this->info['td']->attr_transform_pre['bgcolor'] =
  99         $this->info['th']->attr_transform_pre['bgcolor'] = new HTMLPurifier_AttrTransform_BgColor();
 100
 101         $this->info['img']->attr_transform_pre['border'] = new HTMLPurifier_AttrTransform_Border();
 102
 103         $this->info['img']->attr_transform_pre['name'] =
 104         $this->info['a']->attr_transform_pre['name'] = new HTMLPurifier_AttrTransform_Name();
 105
 106         $this->info['td']->attr_transform_pre['width'] =
 107         $this->info['th']->attr_transform_pre['width'] =
 108         $this->info['hr']->attr_transform_pre['width'] = new HTMLPurifier_AttrTransform_Length('width');
 109
 110         $this->info['td']->attr_transform_pre['nowrap'] =
 111         $this->info['th']->attr_transform_pre['nowrap'] = new HTMLPurifier_AttrTransform_BoolToCSS('nowrap', 'white-space:nowrap;');
 112
 113         $this->info['td']->attr_transform_pre['height'] =
 114         $this->info['th']->attr_transform_pre['height'] = new HTMLPurifier_AttrTransform_Length('height');
 115
 116         $this->info['img']->attr_transform_pre['hspace'] = new HTMLPurifier_AttrTransform_ImgSpace('hspace');
 117         $this->info['img']->attr_transform_pre['vspace'] = new HTMLPurifier_AttrTransform_ImgSpace('vspace');
 118
 119         $this->info['hr']->attr_transform_pre['size'] = new HTMLPurifier_AttrTransform_Length('size', 'height');
 120
 121         // this transformation is not precise but often good enough.
 122         // different browsers use different styles to designate noshade
 123         $this->info['hr']->attr_transform_pre['noshade'] = new HTMLPurifier_AttrTransform_BoolToCSS('noshade', 'color:#808080;background-color:#808080;border: 0;');
 124
 125         $this->info['br']->attr_transform_pre['clear'] =
 126             new HTMLPurifier_AttrTransform_EnumToCSS('clear', array(
 127                 'left' => 'clear:left;',
 128                 'right' => 'clear:right;',
 129                 'all' => 'clear:both;',
 130                 'none' => 'clear:none;',
 131             ));
 132
 133         // this is a slightly unreasonable attribute
 134         $this->info['caption']->attr_transform_pre['align'] =
 135             new HTMLPurifier_AttrTransform_EnumToCSS('align', array(
 136                 // we're following IE's behavior, not Firefox's, due
 137                 // to the fact that no one supports caption-side:right,
 138                 // W3C included (with CSS 2.1)
 139                 'left' => 'text-align:left;',
 140                 'right' => 'text-align:right;',
 141                 'top' => 'caption-side:top;',
 142                 'bottom' => 'caption-side:bottom;' // not supported by IE
 143             ));
 144
 145         $this->info['table']->attr_transform_pre['align'] =
 146             new HTMLPurifier_AttrTransform_EnumToCSS('align', array(
 147                 'left' => 'float:left;',
 148                 'center' => 'margin-left:auto;margin-right:auto;',
 149                 'right' => 'float:right;'
 150             ));
 151
 152         $this->info['img']->attr_transform_pre['align'] =
 153             new HTMLPurifier_AttrTransform_EnumToCSS('align', array(
 154                 'left' => 'float:left;',
 155                 'right' => 'float:right;',
 156                 'top' => 'vertical-align:top;',
 157                 'middle' => 'vertical-align:middle;',
 158                 'bottom' => 'vertical-align:baseline;',
 159             ));
 160
 161         $this->info['hr']->attr_transform_pre['align'] =
 162             new HTMLPurifier_AttrTransform_EnumToCSS('align', array(
 163                 'left' => 'margin-left:0;margin-right:auto;text-align:left;',
 164                 'center' => 'margin-left:auto;margin-right:auto;text-align:center;',
 165                 'right' => 'margin-left:auto;margin-right:0;text-align:right;'
 166             ));
 167
 168         $ul_types = array(
 169             'disc'   => 'list-style-type:disc;',
 170             'square' => 'list-style-type:square;',
 171             'circle' => 'list-style-type:circle;'
 172         );
 173         $ol_types = array(
 174             '1'   => 'list-style-type:decimal;',
 175             'i'   => 'list-style-type:lower-roman;',
 176             'I'   => 'list-style-type:upper-roman;',
 177             'a'   => 'list-style-type:lower-alpha;',
 178             'A'   => 'list-style-type:upper-alpha;'
 179         );
 180         $li_types = $ul_types + $ol_types;
 181
 182         $this->info['ul']->attr_transform_pre['type'] =
 183             new HTMLPurifier_AttrTransform_EnumToCSS('type', $ul_types);
 184         $this->info['ol']->attr_transform_pre['type'] =
 185             new HTMLPurifier_AttrTransform_EnumToCSS('type', $ol_types, true);
 186         $this->info['li']->attr_transform_pre['type'] =
 187             new HTMLPurifier_AttrTransform_EnumToCSS('type', $li_types, true);
 188
 189
 190     }
 191
 192     var $defines_child_def = true;
 193     function getChildDef($def) {
 194         if ($def->content_model_type != 'strictblockquote') return false;
 195         return new HTMLPurifier_ChildDef_StrictBlockquote($def->content_model);
 196     }
 197
 198 }
 199
 200 ?>