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