library/HTMLPurifier/XHTMLDefinition.php

   1 <?php
   2
   3 require_once 'HTMLPurifier/HTMLDefinition.php';
   4
   5 require_once 'HTMLPurifier/AttrTypes.php';
   6 require_once 'HTMLPurifier/AttrCollection.php';
   7
   8 require_once 'HTMLPurifier/HTMLModule.php';
   9 require_once 'HTMLPurifier/HTMLModule/Text.php';
  10 require_once 'HTMLPurifier/HTMLModule/Hypertext.php';
  11 require_once 'HTMLPurifier/HTMLModule/List.php';
  12 require_once 'HTMLPurifier/HTMLModule/Presentation.php';
  13 require_once 'HTMLPurifier/HTMLModule/Edit.php';
  14 require_once 'HTMLPurifier/HTMLModule/Bdo.php';
  15 require_once 'HTMLPurifier/HTMLModule/Tables.php';
  16
  17 /**
  18  * Next-generation HTML definition that will supplant HTMLPurifier_HTMLDefinition
  19  */
  20 class HTMLPurifier_XHTMLDefinition extends HTMLPurifier_HTMLDefinition
  21 {
  22
  23     var $modules = array();
  24     var $attr_types;
  25     var $attr_collection;
  26     var $content_sets;
  27
  28     function HTMLPurifier_XHTMLDefinition($config) {
  29
  30         $this->modules['Text']          = new HTMLPurifier_HTMLModule_Text();
  31         $this->modules['Hypertext']     = new HTMLPurifier_HTMLModule_Hypertext();
  32         $this->modules['List']          = new HTMLPurifier_HTMLModule_List();
  33         $this->modules['Presentation']  = new HTMLPurifier_HTMLModule_Presentation();
  34         $this->modules['Edit']          = new HTMLPurifier_HTMLModule_Edit();
  35         $this->modules['Bdo']           = new HTMLPurifier_HTMLModule_Bdo();
  36         $this->modules['Tables']        = new HTMLPurifier_HTMLModule_Tables();
  37
  38         $this->attr_types = new HTMLPurifier_AttrTypes();
  39         $this->attr_collection = new HTMLPurifier_AttrCollection();
  40
  41     }
  42
  43     function setup($config) {
  44
  45         // perform attribute collection substitutions
  46         $this->attr_collection->setup($this->attr_types, $this->modules);
  47
  48         // populate content_sets based on module hints
  49         $content_sets = array();
  50         foreach ($this->modules as $module_i => $module) {
  51             foreach ($module->content_sets as $key => $value) {
  52                 if (isset($content_sets[$key])) {
  53                     // add it into the existing content set
  54                     $content_sets[$key] = $content_sets[$key] . ' | ' . $value;
  55                 } else {
  56                     $content_sets[$key] = $value;
  57                 }
  58             }
  59         }
  60
  61         // perform content_set expansions
  62         foreach ($content_sets as $i => $set) {
  63             // only performed once, so infinite recursion is not
  64             // a problem, you'll just have a stray $Set lying around
  65             // at the end
  66             $content_sets[$i] =
  67                 str_replace(
  68                     array_keys($content_sets),
  69                     array_values($content_sets),
  70                 $set);
  71         }
  72         // define convenient variables
  73         $content_sets_keys   = array_keys($content_sets);
  74         $content_sets_values = array_values($content_sets);
  75         foreach ($content_sets as $name => $set) {
  76             $this->content_sets[$name] = $this->convertToLookup($set);
  77         }
  78
  79         foreach ($this->modules as $module_i => $module) {
  80             foreach ($module->info as $name => $def) {
  81                 $def =& $this->modules[$module_i]->info[$name];
  82
  83                 // attribute value expansions
  84                 $this->attr_collection->performInclusions($def->attr);
  85                 $this->attr_collection->expandIdentifiers(
  86                     $def->attr, $this->attr_types);
  87
  88                 // perform content model expansions
  89                 $content_model = $def->content_model;
  90                 if (is_string($content_model)) {
  91                     if (strpos($content_model, 'Inline') !== false) {
  92                         $def->descendants_are_inline = true;
  93                     }
  94                     $def->content_model = str_replace(
  95                         $content_sets_keys, $content_sets_values, $content_model);
  96                 }
  97
  98                 // get child def from content model
  99                 $def->child = $this->getChildDef($def);
 100
 101                 // setup info
 102                 $this->info[$name] = $def;
 103                 if ($this->info_parent == $name) {
 104                     $this->info_parent_def = $this->info[$name];
 105                 }
 106             }
 107         }
 108
 109         $this->setupAttrTransform($config);
 110         $this->setupBlockWrapper($config);
 111         $this->setupParent($config);
 112
 113     }
 114
 115     function setupAttrTransform($config) {
 116         $this->info_attr_transform_post[] = new HTMLPurifier_AttrTransform_Lang();
 117     }
 118
 119     function setupBlockWrapper($config) {
 120         $block_wrapper = $config->get('HTML', 'BlockWrapper');
 121         if (isset($this->content_sets['Block'][$block_wrapper])) {
 122             $this->info_block_wrapper = $block_wrapper;
 123         } else {
 124             trigger_error('Cannot use non-block element as block wrapper.',
 125                 E_USER_ERROR);
 126         }
 127     }
 128
 129     function setupParent($config) {
 130         $parent = $config->get('HTML', 'Parent');
 131         if (isset($this->info[$parent])) {
 132             $this->info_parent = $parent;
 133         } else {
 134             trigger_error('Cannot use unrecognized element as parent.',
 135                 E_USER_ERROR);
 136         }
 137         $this->info_parent_def = $this->info[$this->info_parent];
 138     }
 139
 140     function getChildDef($def) {
 141         $value = $def->content_model;
 142         $type  = $def->content_model_type;
 143         switch ($type) {
 144             case 'required':
 145                 return new HTMLPurifier_ChildDef_Required($value);
 146             case 'optional':
 147                 return new HTMLPurifier_ChildDef_Optional($value);
 148             case 'empty':
 149                 return new HTMLPurifier_ChildDef_Empty();
 150             case 'strictblockquote':
 151                 return new HTMLPurifier_ChildDef_StrictBlockquote($value);
 152             case 'table':
 153                 return new HTMLPurifier_ChildDef_Table();
 154             case 'chameleon':
 155                 $value = explode('!', $value);
 156                 return new HTMLPurifier_ChildDef_Chameleon($value[0], $value[1]);
 157             case 'custom':
 158                 return new HTMLPurifier_ChildDef_Custom($value);
 159         }
 160         if ($value) return new HTMLPurifier_ChildDef_Optional($value);
 161         return new HTMLPurifier_ChildDef_Empty();
 162     }
 163
 164     function convertToLookup($string) {
 165         $array = explode('|', str_replace(' ', '', $string));
 166         $ret = array();
 167         foreach ($array as $i => $k) {
 168             $ret[$k] = true;
 169         }
 170         return $ret;
 171     }
 172
 173 }
 174
 175 ?>