plugins/phorum/htmlpurifier.php

   1 <?php
   2
   3 /**
   4  * HTML Purifier Phorum Mod. Filter your HTML the Standards-Compliant Way!
   5  *
   6  * This Phorum mod enables users to post raw HTML into Phorum.  But never
   7  * fear: with the help of HTML Purifier, this HTML will be beat into
   8  * de-XSSed and standards-compliant form, safe for general consumption.
   9  * It is not recommended, but possible to run this mod in parallel
  10  * with other formatters (in short, please DISABLE the BBcode mod).
  11  *
  12  * For help migrating from your previous markup language to pure HTML
  13  * please check the migrate.bbcode.php file.
  14  *
  15  * If you'd like to use this with a WYSIWYG editor, make sure that
  16  * editor sets $PHORUM['mod_htmlpurifier']['wysiwyg'] to true. Otherwise,
  17  * administrators who need to edit other people's comments may be at
  18  * risk for some nasty attacks.
  19  *
  20  * Tested with Phorum 5.2.6.
  21  */
  22
  23 // Note: Cache data is base64 encoded because Phorum insists on flinging
  24 // to the user and expecting it to come back unharmed, newlines and
  25 // all, which ain't happening. It's slower, it takes up more space, but
  26 // at least it won't get mutilated
  27
  28 /**
  29  * Purifies a data array
  30  */
  31 function phorum_htmlpurifier_format($data)
  32 {
  33     $PHORUM = $GLOBALS["PHORUM"];
  34
  35     $purifier =& HTMLPurifier::getInstance();
  36     $cache_serial = $PHORUM['mod_htmlpurifier']['body_cache_serial'];
  37
  38     foreach($data as $message_id => $message){
  39         if(isset($message['body'])) {
  40
  41             if ($message_id) {
  42                 // we're dealing with a real message, not a fake, so
  43                 // there a number of shortcuts that can be taken
  44
  45                 if (isset($message['meta']['htmlpurifier_light'])) {
  46                     // format hook was called outside of Phorum's normal
  47                     // functions, do the abridged purification
  48                     $data[$message_id]['body'] = $purifier->purify($message['body']);
  49                     continue;
  50                 }
  51
  52                 if (!empty($PHORUM['args']['purge'])) {
  53                     // purge the cache, must be below the following if
  54                     unset($message['meta']['body_cache']);
  55                 }
  56
  57                 if (
  58                     isset($message['meta']['body_cache']) &&
  59                     isset($message['meta']['body_cache_serial']) &&
  60                     $message['meta']['body_cache_serial'] == $cache_serial
  61                 ) {
  62                     // cached version is present, bail out early
  63                     $data[$message_id]['body'] = base64_decode($message['meta']['body_cache']);
  64                     continue;
  65                 }
  66             }
  67
  68             // migration might edit this array, that's why it's defined
  69             // so early
  70             $updated_message = array();
  71
  72             // create the $body variable
  73             if (
  74                 $message_id && // message must be real to migrate
  75                 !isset($message['meta']['body_cache_serial'])
  76             ) {
  77                 // perform migration
  78                 $fake_data = array();
  79                 list($signature, $edit_message) = phorum_htmlpurifier_remove_sig_and_editmessage($message);
  80                 $fake_data[$message_id] = $message;
  81                 $fake_data = phorum_htmlpurifier_migrate($fake_data);
  82                 $body = $fake_data[$message_id]['body'];
  83                 $body = str_replace("<phorum break>\n", "\n", $body);
  84                 $updated_message['body'] = $body; // save it in
  85                 $body .= $signature . $edit_message; // add it back in
  86             } else {
  87                 // reverse Phorum's pre-processing
  88                 $body = $message['body'];
  89                 // order is important
  90                 $body = str_replace("<phorum break>\n", "\n", $body);
  91                 $body = str_replace(array('&lt;','&gt;','&amp;', '&quot;'), array('<','>','&','"'), $body);
  92                 if (!$message_id && defined('PHORUM_CONTROL_CENTER')) {
  93                     // we're in control.php, so it was double-escaped
  94                     $body = str_replace(array('&lt;','&gt;','&amp;', '&quot;'), array('<','>','&','"'), $body);
  95                 }
  96             }
  97
  98             $body = $purifier->purify($body);
  99
 100             // dynamically update the cache (MUST BE DONE HERE!)
 101             // this is inefficient because it's one db call per
 102             // cache miss, but once the cache is in place things are
 103             // a lot zippier.
 104
 105             if ($message_id) { // make sure it's not a fake id
 106                 $updated_message['meta'] = $message['meta'];
 107                 $updated_message['meta']['body_cache'] = base64_encode($body);
 108                 $updated_message['meta']['body_cache_serial'] = $cache_serial;
 109                 phorum_db_update_message($message_id, $updated_message);
 110             }
 111
 112             // must not get overloaded until after we cache it, otherwise
 113             // we'll inadvertently change the original text
 114             $data[$message_id]['body'] = $body;
 115
 116         }
 117     }
 118
 119     return $data;
 120 }
 121
 122 // -----------------------------------------------------------------------
 123 // This is fragile code, copied from read.php:596 (Phorum 5.2.6). Please
 124 // keep this code in-sync with Phorum
 125
 126 /**
 127  * Generates a signature based on a message array
 128  */
 129 function phorum_htmlpurifier_generate_sig($row) {
 130     $phorum_sig = '';
 131     if(isset($row["user"]["signature"])
 132        && isset($row['meta']['show_signature']) && $row['meta']['show_signature']==1){
 133            $phorum_sig=trim($row["user"]["signature"]);
 134            if(!empty($phorum_sig)){
 135                $phorum_sig="\n\n$phorum_sig";
 136            }
 137     }
 138     return $phorum_sig;
 139 }
 140
 141 /**
 142  * Generates an edit message based on a message array
 143  */
 144 function phorum_htmlpurifier_generate_editmessage($row) {
 145     $PHORUM = $GLOBALS['PHORUM'];
 146     $editmessage = '';
 147     if(isset($row['meta']['edit_count']) && $row['meta']['edit_count'] > 0) {
 148         $editmessage = str_replace ("%count%", $row['meta']['edit_count'], $PHORUM["DATA"]["LANG"]["EditedMessage"]);
 149         $editmessage = str_replace ("%lastedit%", phorum_date($PHORUM["short_date_time"],$row['meta']['edit_date']),  $editmessage);
 150         $editmessage = str_replace ("%lastuser%", $row['meta']['edit_username'],  $editmessage);
 151         $editmessage = "\n\n\n\n$editmessage";
 152     }
 153     return $editmessage;
 154 }
 155
 156 // End fragile code
 157 // -----------------------------------------------------------------------
 158
 159 /**
 160  * Removes the signature and edit message from a message
 161  * @param $row Message passed by reference
 162  */
 163 function phorum_htmlpurifier_remove_sig_and_editmessage(&$row) {
 164     $signature = phorum_htmlpurifier_generate_sig($row);
 165     $editmessage = phorum_htmlpurifier_generate_editmessage($row);
 166     $replacements = array();
 167     // we need to remove add <phorum break> as that is the form these
 168     // extra bits are in.
 169     if ($signature) $replacements[str_replace("\n", "<phorum break>\n", $signature)] = '';
 170     if ($editmessage) $replacements[str_replace("\n", "<phorum break>\n", $editmessage)] = '';
 171     $row['body'] = strtr($row['body'], $replacements);
 172     return array($signature, $editmessage);
 173 }
 174
 175 /**
 176  * Indicate that data is fully HTML and not from migration, invalidate
 177  * previous caches
 178  * @note This function could generate the actual cache entries, but
 179  *       since there's data missing that must be deferred to the first read
 180  */
 181 function phorum_htmlpurifier_posting($message) {
 182     $PHORUM = $GLOBALS["PHORUM"];
 183     unset($message['meta']['body_cache']); // invalidate the cache
 184     $message['meta']['body_cache_serial'] = $PHORUM['mod_htmlpurifier']['body_cache_serial'];
 185     return $message;
 186 }
 187
 188 /**
 189  * Overload quoting mechanism to prevent default, mail-style quote from happening
 190  */
 191 function phorum_htmlpurifier_quote($array) {
 192     $PHORUM = $GLOBALS["PHORUM"];
 193     $purifier =& HTMLPurifier::getInstance();
 194     $text = $purifier->purify($array[1]);
 195     $source = htmlspecialchars($array[0]);
 196     return "<blockquote cite=\"$source\">\n$text\n</blockquote>";
 197 }
 198
 199 /**
 200  * Ensure that our format hook is processed last. Also, loads the library.
 201  * @credits <http://secretsauce.phorum.org/snippets/make_bbcode_last_formatter.php.txt>
 202  */
 203 function phorum_htmlpurifier_common() {
 204
 205     require_once(dirname(__FILE__).'/htmlpurifier/HTMLPurifier.auto.php');
 206     require(dirname(__FILE__).'/init-config.php');
 207
 208     $config = phorum_htmlpurifier_get_config();
 209     HTMLPurifier::getInstance($config);
 210
 211     // increment revision.txt if you want to invalidate the cache
 212     $GLOBALS['PHORUM']['mod_htmlpurifier']['body_cache_serial'] = $config->getSerial();
 213
 214     // load migration
 215     if (file_exists(dirname(__FILE__) . '/migrate.php')) {
 216         include(dirname(__FILE__) . '/migrate.php');
 217     } else {
 218         echo '<strong>Error:</strong> No migration path specified for HTML Purifier, please check
 219         <tt>modes/htmlpurifier/migrate.bbcode.php</tt> for instructions on
 220         how to migrate from your previous markup language.';
 221         exit;
 222     }
 223
 224     if (!function_exists('phorum_htmlpurifier_migrate')) {
 225         // Dummy function
 226         function phorum_htmlpurifier_migrate($data) {return $data;}
 227     }
 228
 229 }
 230
 231 /**
 232  * Pre-emptively performs purification if it looks like a WYSIWYG editor
 233  * is being used
 234  */
 235 function phorum_htmlpurifier_before_editor($message) {
 236     if (!empty($GLOBALS['PHORUM']['mod_htmlpurifier']['wysiwyg'])) {
 237         if (!empty($message['body'])) {
 238             $body = $message['body'];
 239             // de-entity-ize contents
 240             $body = str_replace(array('&lt;','&gt;','&amp;'), array('<','>','&'), $body);
 241             $purifier =& HTMLPurifier::getInstance();
 242             $body = $purifier->purify($body);
 243             // re-entity-ize contents
 244             $body = htmlspecialchars($body, ENT_QUOTES, $GLOBALS['PHORUM']['DATA']['CHARSET']);
 245             $message['body'] = $body;
 246         }
 247     }
 248     return $message;
 249 }
 250
 251 function phorum_htmlpurifier_editor_after_subject() {
 252     // don't show this message if it's a WYSIWYG editor, since it will
 253     // then be handled automatically
 254     if (!empty($GLOBALS['PHORUM']['mod_htmlpurifier']['wysiwyg'])) {
 255         $i = $GLOBALS['PHORUM']['DATA']['MODE'];
 256         if ($i == 'quote' || $i == 'edit' || $i == 'moderation') {
 257           ?>
 258           <div>
 259             <p>
 260               <strong>Notice:</strong> HTML has been scrubbed for your safety.
 261               If you would like to see the original, turn off WYSIWYG mode
 262               (consult your administrator for details.)
 263             </p>
 264           </div>
 265           <?php
 266         }
 267         return;
 268     }
 269     if (!empty($GLOBALS['PHORUM']['mod_htmlpurifier']['suppress_message'])) return;
 270     ?><div class="htmlpurifier-help">
 271     <p>
 272         <strong>HTML input</strong> is enabled. Make sure you escape all HTML and
 273         angled brackets with <code>&amp;lt;</code> and <code>&amp;gt;</code>.
 274     </p><?php
 275             $purifier =& HTMLPurifier::getInstance();
 276             $config = $purifier->config;
 277             if ($config->get('AutoFormat.AutoParagraph')) {
 278                 ?><p>
 279                     <strong>Auto-paragraphing</strong> is enabled. Double
 280                     newlines will be converted to paragraphs; for single
 281                     newlines, use the <code>pre</code> tag.
 282                 </p><?php
 283             }
 284             $html_definition = $config->getDefinition('HTML');
 285             $allowed = array();
 286             foreach ($html_definition->info as $name => $x) $allowed[] = "<code>$name</code>";
 287             sort($allowed);
 288             $allowed_text = implode(', ', $allowed);
 289             ?><p><strong>Allowed tags:</strong> <?php
 290             echo $allowed_text;
 291             ?>.</p><?php
 292         ?>
 293     </p>
 294     <p>
 295         For inputting literal code such as HTML and PHP for display, use
 296         CDATA tags to auto-escape your angled brackets, and <code>pre</code>
 297         to preserve newlines:
 298     </p>
 299     <pre>&lt;pre&gt;&lt;![CDATA[
 300 <em>Place code here</em>
 301 ]]&gt;&lt;/pre&gt;</pre>
 302     <p>
 303         Power users, you can hide this notice with:
 304         <pre>.htmlpurifier-help {display:none;}</pre>
 305     </p>
 306     </div><?php
 307 }
 308
 309 // vim: et sw=4 sts=4