library/HTMLPurifier/Generator.php

   1 <?php
   2
   3 /**
   4  * Generates HTML from tokens.
   5  * @todo Refactor interface so that configuration/context is determined
   6  *       upon instantiation, no need for messy generateFromTokens() calls
   7  * @todo Make some of the more internal functions protected, and have
   8  *       unit tests work around that
   9  */
  10 class HTMLPurifier_Generator
  11 {
  12
  13     /**
  14      * Whether or not generator should produce XML output
  15      */
  16     private $_xhtml = true;
  17
  18     /**
  19      * :HACK: Whether or not generator should comment the insides of <script> tags
  20      */
  21     private $_scriptFix = false;
  22
  23     /**
  24      * Cache of HTMLDefinition during HTML output to determine whether or
  25      * not attributes should be minimized.
  26      */
  27     private $_def;
  28
  29     /**
  30      * Cache of %Output.SortAttr
  31      */
  32     private $_sortAttr;
  33
  34     /**
  35      * Cache of %Output.FlashCompat
  36      */
  37     private $_flashCompat;
  38
  39     /**
  40      * Stack for keeping track of object information when outputting IE
  41      * compatibility code.
  42      */
  43     private $_flashStack = array();
  44
  45     /**
  46      * Configuration for the generator
  47      */
  48     protected $config;
  49
  50     /**
  51      * @param $config Instance of HTMLPurifier_Config
  52      * @param $context Instance of HTMLPurifier_Context
  53      */
  54     public function __construct($config, $context) {
  55         $this->config = $config;
  56         $this->_scriptFix = $config->get('Output.CommentScriptContents');
  57         $this->_sortAttr = $config->get('Output.SortAttr');
  58         $this->_flashCompat = $config->get('Output.FlashCompat');
  59         $this->_def = $config->getHTMLDefinition();
  60         $this->_xhtml = $this->_def->doctype->xml;
  61     }
  62
  63     /**
  64      * Generates HTML from an array of tokens.
  65      * @param $tokens Array of HTMLPurifier_Token
  66      * @param $config HTMLPurifier_Config object
  67      * @return Generated HTML
  68      */
  69     public function generateFromTokens($tokens) {
  70         if (!$tokens) return '';
  71
  72         // Basic algorithm
  73         $html = '';
  74         for ($i = 0, $size = count($tokens); $i < $size; $i++) {
  75             if ($this->_scriptFix && $tokens[$i]->name === 'script'
  76                 && $i + 2 < $size && $tokens[$i+2] instanceof HTMLPurifier_Token_End) {
  77                 // script special case
  78                 // the contents of the script block must be ONE token
  79                 // for this to work.
  80                 $html .= $this->generateFromToken($tokens[$i++]);
  81                 $html .= $this->generateScriptFromToken($tokens[$i++]);
  82             }
  83             $html .= $this->generateFromToken($tokens[$i]);
  84         }
  85
  86         // Tidy cleanup
  87         if (extension_loaded('tidy') && $this->config->get('Output.TidyFormat')) {
  88             $tidy = new Tidy;
  89             $tidy->parseString($html, array(
  90                'indent'=> true,
  91                'output-xhtml' => $this->_xhtml,
  92                'show-body-only' => true,
  93                'indent-spaces' => 2,
  94                'wrap' => 68,
  95             ), 'utf8');
  96             $tidy->cleanRepair();
  97             $html = (string) $tidy; // explicit cast necessary
  98         }
  99
 100         // Normalize newlines to system defined value
 101         $nl = $this->config->get('Output.Newline');
 102         if ($nl === null) $nl = PHP_EOL;
 103         if ($nl !== "\n") $html = str_replace("\n", $nl, $html);
 104         return $html;
 105     }
 106
 107     /**
 108      * Generates HTML from a single token.
 109      * @param $token HTMLPurifier_Token object.
 110      * @return Generated HTML
 111      */
 112     public function generateFromToken($token) {
 113         if (!$token instanceof HTMLPurifier_Token) {
 114             trigger_error('Cannot generate HTML from non-HTMLPurifier_Token object', E_USER_WARNING);
 115             return '';
 116
 117         } elseif ($token instanceof HTMLPurifier_Token_Start) {
 118             $attr = $this->generateAttributes($token->attr, $token->name);
 119             if ($this->_flashCompat) {
 120                 if ($token->name == "object") {
 121                     $flash = new stdclass();
 122                     $flash->attr = $token->attr;
 123                     $flash->param = array();
 124                     $this->_flashStack[] = $flash;
 125                 }
 126             }
 127             return '<' . $token->name . ($attr ? ' ' : '') . $attr . '>';
 128
 129         } elseif ($token instanceof HTMLPurifier_Token_End) {
 130             $_extra = '';
 131             if ($this->_flashCompat) {
 132                 if ($token->name == "object" && !empty($this->_flashStack)) {
 133                     $flash = array_pop($this->_flashStack);
 134                     $compat_token = new HTMLPurifier_Token_Empty("embed");
 135                     foreach ($flash->attr as $name => $val) {
 136                         if ($name == "classid") continue;
 137                         if ($name == "type") continue;
 138                         if ($name == "data") $name = "src";
 139                         $compat_token->attr[$name] = $val;
 140                     }
 141                     foreach ($flash->param as $name => $val) {
 142                         if ($name == "movie") $name = "src";
 143                         $compat_token->attr[$name] = $val;
 144                     }
 145                     $_extra = "<!--[if IE]>".$this->generateFromToken($compat_token)."<![endif]-->";
 146                 }
 147             }
 148             return $_extra . '</' . $token->name . '>';
 149
 150         } elseif ($token instanceof HTMLPurifier_Token_Empty) {
 151             if ($this->_flashCompat && $token->name == "param" && !empty($this->_flashStack)) {
 152                 $this->_flashStack[count($this->_flashStack)-1]->param[$token->attr['name']] = $token->attr['value'];
 153             }
 154             $attr = $this->generateAttributes($token->attr, $token->name);
 155              return '<' . $token->name . ($attr ? ' ' : '') . $attr .
 156                 ( $this->_xhtml ? ' /': '' ) // <br /> v. <br>
 157                 . '>';
 158
 159         } elseif ($token instanceof HTMLPurifier_Token_Text) {
 160             return $this->escape($token->data, ENT_NOQUOTES);
 161
 162         } elseif ($token instanceof HTMLPurifier_Token_Comment) {
 163             return '<!--' . $token->data . '-->';
 164         } else {
 165             return '';
 166
 167         }
 168     }
 169
 170     /**
 171      * Special case processor for the contents of script tags
 172      * @warning This runs into problems if there's already a literal
 173      *          --> somewhere inside the script contents.
 174      */
 175     public function generateScriptFromToken($token) {
 176         if (!$token instanceof HTMLPurifier_Token_Text) return $this->generateFromToken($token);
 177         // Thanks <http://lachy.id.au/log/2005/05/script-comments>
 178         $data = preg_replace('#//\s*$#', '', $token->data);
 179         return '<!--//--><![CDATA[//><!--' . "\n" . trim($data) . "\n" . '//--><!]]>';
 180     }
 181
 182     /**
 183      * Generates attribute declarations from attribute array.
 184      * @note This does not include the leading or trailing space.
 185      * @param $assoc_array_of_attributes Attribute array
 186      * @param $element Name of element attributes are for, used to check
 187      *        attribute minimization.
 188      * @return Generate HTML fragment for insertion.
 189      */
 190     public function generateAttributes($assoc_array_of_attributes, $element = false) {
 191         $html = '';
 192         if ($this->_sortAttr) ksort($assoc_array_of_attributes);
 193         foreach ($assoc_array_of_attributes as $key => $value) {
 194             if (!$this->_xhtml) {
 195                 // Remove namespaced attributes
 196                 if (strpos($key, ':') !== false) continue;
 197                 // Check if we should minimize the attribute: val="val" -> val
 198                 if ($element && !empty($this->_def->info[$element]->attr[$key]->minimized)) {
 199                     $html .= $key . ' ';
 200                     continue;
 201                 }
 202             }
 203             $html .= $key.'="'.$this->escape($value).'" ';
 204         }
 205         return rtrim($html);
 206     }
 207
 208     /**
 209      * Escapes raw text data.
 210      * @todo This really ought to be protected, but until we have a facility
 211      *       for properly generating HTML here w/o using tokens, it stays
 212      *       public.
 213      * @param $string String data to escape for HTML.
 214      * @param $quote Quoting style, like htmlspecialchars. ENT_NOQUOTES is
 215      *               permissible for non-attribute output.
 216      * @return String escaped data.
 217      */
 218     public function escape($string, $quote = null) {
 219         // Workaround for APC bug on Mac Leopard reported by sidepodcast
 220         // http://htmlpurifier.org/phorum/read.php?3,4823,4846
 221         if ($quote === null) $quote = ENT_COMPAT;
 222         return htmlspecialchars($string, $quote, 'UTF-8');
 223     }
 224
 225 }
 226
 227 // vim: et sw=4 sts=4