library/HTMLPurifier/ContentSets.php

   1 <?php
   2
   3 /**
   4  * @todo Unit test
   5  */
   6 class HTMLPurifier_ContentSets
   7 {
   8
   9     /**
  10      * List of content set strings (pipe separators) indexed by name.
  11      * @type array
  12      */
  13     public $info = array();
  14
  15     /**
  16      * List of content set lookups (element => true) indexed by name.
  17      * @type array
  18      * @note This is in HTMLPurifier_HTMLDefinition->info_content_sets
  19      */
  20     public $lookup = array();
  21
  22     /**
  23      * Synchronized list of defined content sets (keys of info).
  24      * @type array
  25      */
  26     protected $keys = array();
  27     /**
  28      * Synchronized list of defined content values (values of info).
  29      * @type array
  30      */
  31     protected $values = array();
  32
  33     /**
  34      * Merges in module's content sets, expands identifiers in the content
  35      * sets and populates the keys, values and lookup member variables.
  36      * @param HTMLPurifier_HTMLModule[] $modules List of HTMLPurifier_HTMLModule
  37      */
  38     public function __construct($modules)
  39     {
  40         if (!is_array($modules)) {
  41             $modules = array($modules);
  42         }
  43         // populate content_sets based on module hints
  44         // sorry, no way of overloading
  45         foreach ($modules as $module) {
  46             foreach ($module->content_sets as $key => $value) {
  47                 $temp = $this->convertToLookup($value);
  48                 if (isset($this->lookup[$key])) {
  49                     // add it into the existing content set
  50                     $this->lookup[$key] = array_merge($this->lookup[$key], $temp);
  51                 } else {
  52                     $this->lookup[$key] = $temp;
  53                 }
  54             }
  55         }
  56         $old_lookup = false;
  57         while ($old_lookup !== $this->lookup) {
  58             $old_lookup = $this->lookup;
  59             foreach ($this->lookup as $i => $set) {
  60                 $add = array();
  61                 foreach ($set as $element => $x) {
  62                     if (isset($this->lookup[$element])) {
  63                         $add += $this->lookup[$element];
  64                         unset($this->lookup[$i][$element]);
  65                     }
  66                 }
  67                 $this->lookup[$i] += $add;
  68             }
  69         }
  70
  71         foreach ($this->lookup as $key => $lookup) {
  72             $this->info[$key] = implode(' | ', array_keys($lookup));
  73         }
  74         $this->keys   = array_keys($this->info);
  75         $this->values = array_values($this->info);
  76     }
  77
  78     /**
  79      * Accepts a definition; generates and assigns a ChildDef for it
  80      * @param HTMLPurifier_ElementDef $def HTMLPurifier_ElementDef reference
  81      * @param HTMLPurifier_HTMLModule $module Module that defined the ElementDef
  82      */
  83     public function generateChildDef(&$def, $module)
  84     {
  85         if (!empty($def->child)) { // already done!
  86             return;
  87         }
  88         $content_model = $def->content_model;
  89         if (is_string($content_model)) {
  90             // Assume that $this->keys is alphanumeric
  91             $def->content_model = preg_replace_callback(
  92                 '/\b(' . implode('|', $this->keys) . ')\b/',
  93                 array($this, 'generateChildDefCallback'),
  94                 $content_model
  95             );
  96             //$def->content_model = str_replace(
  97             //    $this->keys, $this->values, $content_model);
  98         }
  99         $def->child = $this->getChildDef($def, $module);
 100     }
 101
 102     public function generateChildDefCallback($matches)
 103     {
 104         return $this->info[$matches[0]];
 105     }
 106
 107     /**
 108      * Instantiates a ChildDef based on content_model and content_model_type
 109      * member variables in HTMLPurifier_ElementDef
 110      * @note This will also defer to modules for custom HTMLPurifier_ChildDef
 111      *       subclasses that need content set expansion
 112      * @param HTMLPurifier_ElementDef $def HTMLPurifier_ElementDef to have ChildDef extracted
 113      * @param HTMLPurifier_HTMLModule $module Module that defined the ElementDef
 114      * @return HTMLPurifier_ChildDef corresponding to ElementDef
 115      */
 116     public function getChildDef($def, $module)
 117     {
 118         $value = $def->content_model;
 119         if (is_object($value)) {
 120             trigger_error(
 121                 'Literal object child definitions should be stored in '.
 122                 'ElementDef->child not ElementDef->content_model',
 123                 E_USER_NOTICE
 124             );
 125             return $value;
 126         }
 127         switch ($def->content_model_type) {
 128             case 'required':
 129                 return new HTMLPurifier_ChildDef_Required($value);
 130             case 'optional':
 131                 return new HTMLPurifier_ChildDef_Optional($value);
 132             case 'empty':
 133                 return new HTMLPurifier_ChildDef_Empty();
 134             case 'custom':
 135                 return new HTMLPurifier_ChildDef_Custom($value);
 136         }
 137         // defer to its module
 138         $return = false;
 139         if ($module->defines_child_def) { // save a func call
 140             $return = $module->getChildDef($def);
 141         }
 142         if ($return !== false) {
 143             return $return;
 144         }
 145         // error-out
 146         trigger_error(
 147             'Could not determine which ChildDef class to instantiate',
 148             E_USER_ERROR
 149         );
 150         return false;
 151     }
 152
 153     /**
 154      * Converts a string list of elements separated by pipes into
 155      * a lookup array.
 156      * @param string $string List of elements
 157      * @return array Lookup array of elements
 158      */
 159     protected function convertToLookup($string)
 160     {
 161         $array = explode('|', str_replace(' ', '', $string));
 162         $ret = array();
 163         foreach ($array as $k) {
 164             $ret[$k] = true;
 165         }
 166         return $ret;
 167     }
 168 }
 169
 170 // vim: et sw=4 sts=4