library/HTMLPurifier/HTMLModule/Tidy.php

   1 <?php
   2
   3 require_once 'HTMLPurifier/HTMLModule.php';
   4
   5 HTMLPurifier_ConfigSchema::define(
   6     'HTML', 'TidyLevel', 'medium', 'string', '
   7 <p>General level of cleanliness the Tidy module should enforce.
   8 There are four allowed values:</p>
   9 <dl>
  10     <dt>none</dt>
  11     <dd>No extra tidying should be done</dd>
  12     <dt>light</dt>
  13     <dd>Only fix elements that would be discarded otherwise due to
  14     lack of support in doctype</dd>
  15     <dt>medium</dt>
  16     <dd>Enforce best practices</dd>
  17     <dt>heavy</dt>
  18     <dd>Transform all deprecated elements and attributes to standards
  19     compliant equivalents</dd>
  20 </dl>
  21 <p>This directive has been available since 2.0.0</p>
  22 ' );
  23 HTMLPurifier_ConfigSchema::defineAllowedValues(
  24     'HTML', 'TidyLevel', array('none', 'light', 'medium', 'heavy')
  25 );
  26
  27 HTMLPurifier_ConfigSchema::define(
  28     'HTML', 'TidyAdd', array(), 'lookup', '
  29 Fixes to add to the default set of Tidy fixes as per your level. This
  30 directive has been available since 2.0.0.
  31 ' );
  32
  33 HTMLPurifier_ConfigSchema::define(
  34     'HTML', 'TidyRemove', array(), 'lookup', '
  35 Fixes to remove from the default set of Tidy fixes as per your level. This
  36 directive has been available since 2.0.0.
  37 ' );
  38
  39 /**
  40  * Abstract class for a set of proprietary modules that clean up (tidy)
  41  * poorly written HTML.
  42  */
  43 class HTMLPurifier_HTMLModule_Tidy extends HTMLPurifier_HTMLModule
  44 {
  45
  46     /**
  47      * List of supported levels. Index zero is a special case "no fixes"
  48      * level.
  49      */
  50     var $levels = array(0 => 'none', 'light', 'medium', 'heavy');
  51
  52     /**
  53      * Default level to place all fixes in. Disabled by default
  54      */
  55     var $defaultLevel = null;
  56
  57     /**
  58      * Lists of fixes used by getFixesForLevel(). Format is:
  59      *      HTMLModule_Tidy->fixesForLevel[$level] = array('fix-1', 'fix-2');
  60      */
  61     var $fixesForLevel = array(
  62         'light'  => array(),
  63         'medium' => array(),
  64         'heavy'  => array()
  65     );
  66
  67     /**
  68      * Lazy load constructs the module by determining the necessary
  69      * fixes to create and then delegating to the populate() function.
  70      * @todo Wildcard matching and error reporting when an added or
  71      *       subtracted fix has no effect.
  72      */
  73     function construct($config) {
  74
  75         // create fixes, initialize fixesForLevel
  76         $fixes = $this->makeFixes();
  77         $this->makeFixesForLevel($fixes);
  78
  79         // figure out which fixes to use
  80         $level = $config->get('HTML', 'TidyLevel');
  81         $fixes_lookup = $this->getFixesForLevel($level);
  82
  83         // get custom fix declarations: these need namespace processing
  84         $add_fixes    = $config->get('HTML', 'TidyAdd');
  85         $remove_fixes = $config->get('HTML', 'TidyRemove');
  86
  87         foreach ($fixes as $name => $fix) {
  88             // needs to be refactored a little to implement globbing
  89             if (
  90                 isset($remove_fixes[$name]) ||
  91                 (!isset($add_fixes[$name]) && !isset($fixes_lookup[$name]))
  92             ) {
  93                 unset($fixes[$name]);
  94             }
  95         }
  96
  97         // populate this module with necessary fixes
  98         $this->populate($fixes);
  99
 100     }
 101
 102     /**
 103      * Retrieves all fixes per a level, returning fixes for that specific
 104      * level as well as all levels below it.
 105      * @param $level String level identifier, see $levels for valid values
 106      * @return Lookup up table of fixes
 107      */
 108     function getFixesForLevel($level) {
 109         if ($level == $this->levels[0]) {
 110             return array();
 111         }
 112         $activated_levels = array();
 113         for ($i = 1, $c = count($this->levels); $i < $c; $i++) {
 114             $activated_levels[] = $this->levels[$i];
 115             if ($this->levels[$i] == $level) break;
 116         }
 117         if ($i == $c) {
 118             trigger_error(
 119                 'Tidy level ' . htmlspecialchars($level) . ' not recognized',
 120                 E_USER_WARNING
 121             );
 122             return array();
 123         }
 124         $ret = array();
 125         foreach ($activated_levels as $level) {
 126             foreach ($this->fixesForLevel[$level] as $fix) {
 127                 $ret[$fix] = true;
 128             }
 129         }
 130         return $ret;
 131     }
 132
 133     /**
 134      * Dynamically populates the $fixesForLevel member variable using
 135      * the fixes array. It may be custom overloaded, used in conjunction
 136      * with $defaultLevel, or not used at all.
 137      */
 138     function makeFixesForLevel($fixes) {
 139         if (!isset($this->defaultLevel)) return;
 140         if (!isset($this->fixesForLevel[$this->defaultLevel])) {
 141             trigger_error(
 142                 'Default level ' . $this->defaultLevel . ' does not exist',
 143                 E_USER_ERROR
 144             );
 145             return;
 146         }
 147         $this->fixesForLevel[$this->defaultLevel] = array_keys($fixes);
 148     }
 149
 150     /**
 151      * Populates the module with transforms and other special-case code
 152      * based on a list of fixes passed to it
 153      * @param $lookup Lookup table of fixes to activate
 154      */
 155     function populate($fixes) {
 156         foreach ($fixes as $name => $fix) {
 157             // determine what the fix is for
 158             list($type, $params) = $this->getFixType($name);
 159             switch ($type) {
 160                 case 'attr_transform_pre':
 161                 case 'attr_transform_post':
 162                     $attr = $params['attr'];
 163                     if (isset($params['element'])) {
 164                         $element = $params['element'];
 165                         if (empty($this->info[$element])) {
 166                             $e =& $this->addBlankElement($element);
 167                         } else {
 168                             $e =& $this->info[$element];
 169                         }
 170                     } else {
 171                         $type = "info_$type";
 172                         $e =& $this;
 173                     }
 174                     $f =& $e->$type;
 175                     $f[$attr] = $fix;
 176                     break;
 177                 case 'tag_transform':
 178                     $this->info_tag_transform[$params['element']] = $fix;
 179                     break;
 180                 case 'child':
 181                 case 'content_model_type':
 182                     $element = $params['element'];
 183                     if (empty($this->info[$element])) {
 184                         $e =& $this->addBlankElement($element);
 185                     } else {
 186                         $e =& $this->info[$element];
 187                     }
 188                     $e->$type = $fix;
 189                     break;
 190                 default:
 191                     trigger_error("Fix type $type not supported", E_USER_ERROR);
 192                     break;
 193             }
 194         }
 195     }
 196
 197     /**
 198      * Parses a fix name and determines what kind of fix it is, as well
 199      * as other information defined by the fix
 200      * @param $name String name of fix
 201      * @return array(string $fix_type, array $fix_parameters)
 202      * @note $fix_parameters is type dependant, see populate() for usage
 203      *       of these parameters
 204      */
 205     function getFixType($name) {
 206         // parse it
 207         $property = $attr = null;
 208         if (strpos($name, '#') !== false) list($name, $property) = explode('#', $name);
 209         if (strpos($name, '@') !== false) list($name, $attr)     = explode('@', $name);
 210
 211         // figure out the parameters
 212         $params = array();
 213         if ($name !== '')    $params['element'] = $name;
 214         if (!is_null($attr)) $params['attr'] = $attr;
 215
 216         // special case: attribute transform
 217         if (!is_null($attr)) {
 218             if (is_null($property)) $property = 'pre';
 219             $type = 'attr_transform_' . $property;
 220             return array($type, $params);
 221         }
 222
 223         // special case: tag transform
 224         if (is_null($property)) {
 225             return array('tag_transform', $params);
 226         }
 227
 228         return array($property, $params);
 229
 230     }
 231
 232     /**
 233      * Defines all fixes the module will perform in a compact
 234      * associative array of fix name to fix implementation.
 235      * @abstract
 236      */
 237     function makeFixes() {}
 238
 239 }
 240
 241