wcfsetup/install/files/lib/system/bbcode/SimpleMessageParser.class.php

   1 <?php
   2 namespace wcf\system\bbcode;
   3 use wcf\data\smiley\Smiley;
   4 use wcf\data\smiley\SmileyCache;
   5 use wcf\system\event\EventHandler;
   6 use wcf\system\SingletonFactory;
   7 use wcf\util\StringUtil;
   8
   9 /**
  10  * Parses urls and smilies in simple messages.
  11  *
  12  * @author      Marcel Werk
  13  * @copyright   2001-2016 WoltLab GmbH
  14  * @license     GNU Lesser General Public License <http://opensource.org/licenses/lgpl-license.php>
  15  * @package     com.woltlab.wcf
  16  * @subpackage  system.bbcode
  17  * @category    Community Framework
  18  */
  19 class SimpleMessageParser extends SingletonFactory {
  20         /**
  21          * forbidden characters
  22          * @var string
  23          */
  24         protected static $illegalChars = '[^\x0-\x2C\x2E\x2F\x3A-\x40\x5B-\x60\x7B-\x7F]+';
  25
  26         /**
  27          * list of smilies
  28          * @var Smiley[]
  29          */
  30         protected $smilies = [];
  31
  32         /**
  33          * cached URLs
  34          * @var string[]
  35          */
  36         protected $cachedURLs = [];
  37
  38         /**
  39          * cached e-mails
  40          * @var string[]
  41          */
  42         protected $cachedEmails = [];
  43
  44         /**
  45          * currently parsed message
  46          * @var string
  47          */
  48         public $message = '';
  49
  50         /**
  51          * @inheritDoc
  52          */
  53         protected function init() {
  54                 parent::init();
  55
  56                 if (MODULE_SMILEY == 1) {
  57                         // get smilies
  58                         $smilies = SmileyCache::getInstance()->getSmilies();
  59                         $categories = SmileyCache::getInstance()->getCategories();
  60                         foreach ($smilies as $categoryID => $categorySmilies) {
  61                                 if ($categories[$categoryID ?: null]->isDisabled) continue;
  62
  63                                 /** @var Smiley $smiley */
  64                                 foreach ($categorySmilies as $smiley) {
  65                                         foreach ($smiley->smileyCodes as $smileyCode) {
  66                                                 $this->smilies[$smileyCode] = '<img src="'.$smiley->getURL().'" alt="'.StringUtil::encodeHTML($smiley->smileyCode).'" />';
  67                                         }
  68                                 }
  69                         }
  70                         krsort($this->smilies);
  71                 }
  72         }
  73
  74         /**
  75          * Parses the given message and returns the parsed message.
  76          *
  77          * @param       string          $message
  78          * @param       boolean         $parseURLs
  79          * @param       boolean         $parseSmilies
  80          * @return      string
  81          */
  82         public function parse($message, $parseURLs = true, $parseSmilies = true) {
  83                 $this->message = $message;
  84                 $this->cachedURLs = $this->cachedEmails = [];
  85
  86                 // call event
  87                 EventHandler::getInstance()->fireAction($this, 'beforeParsing');
  88
  89                 // parse urls
  90                 if ($parseURLs) {
  91                         $this->message = $this->parseURLs($this->message);
  92                 }
  93
  94                 // encode html
  95                 $this->message = StringUtil::encodeHTML($this->message);
  96
  97                 // converts newlines to <br />'s
  98                 $this->message = nl2br($this->message);
  99
 100                 // parse urls
 101                 if ($parseURLs) {
 102                         $this->message = $this->insertCachedURLs($this->message);
 103                 }
 104
 105                 // parse smilies
 106                 if ($parseSmilies) {
 107                         $this->message = $this->parseSmilies($this->message);
 108                 }
 109
 110                 // replace bad html tags (script etc.)
 111                 $badSearch = ['/(javascript):/i', '/(about):/i', '/(vbscript):/i'];
 112                 $badReplace = ['$1<b></b>:', '$1<b></b>:', '$1<b></b>:'];
 113                 $this->message = preg_replace($badSearch, $badReplace, $this->message);
 114
 115                 // call event
 116                 EventHandler::getInstance()->fireAction($this, 'afterParsing');
 117
 118                 return $this->message;
 119         }
 120
 121         /**
 122          * Parses urls.
 123          *
 124          * @param       string          $text
 125          * @return      string          text
 126          */
 127         public function parseURLs($text) {
 128                 // define pattern
 129                 $urlPattern = '~(?<!\B|"|\'|=|/|\]|,|\?)
 130                         (?:                                             # hostname
 131                                 (?:ftp|https?)://'.static::$illegalChars.'(?:\.'.static::$illegalChars.')*
 132                                 |
 133                                 www\.(?:'.static::$illegalChars.'\.)+
 134                                 (?:[a-z]{2,63}(?=\b))                   # tld
 135                         )
 136
 137                         (?::\d+)?                                       # port
 138
 139                         (?:
 140                                 /
 141                                 [^!.,?;"\'<>()\[\]{}\s]*
 142                                 (?:
 143                                         [!.,?;(){}]+ [^!.,?;"\'<>()\[\]{}\s]+
 144                                 )*
 145                         )?
 146                         ~ix';
 147                 $emailPattern = '~(?<!\B|"|\'|=|/|\]|,|:)
 148                         (?:)
 149                         \w+(?:[\.\-]\w+)*
 150                         @
 151                         (?:'.static::$illegalChars.'\.)+                # hostname
 152                         (?:[a-z]{2,4}(?=\b))
 153                         (?!"|\'|\[|\-)
 154                         ~ix';
 155
 156                 // parse urls
 157                 $text = preg_replace_callback($urlPattern, [$this, 'cacheURLsCallback'], $text);
 158
 159                 // parse emails
 160                 if (mb_strpos($text, '@') !== false) {
 161                         $text = preg_replace_callback($emailPattern, [$this, 'cacheEmailsCallback'], $text);
 162                 }
 163
 164                 return $text;
 165         }
 166
 167         /**
 168          * Returns the hash for an matched URL in the message.
 169          *
 170          * @param       array           $matches
 171          * @return      string
 172          */
 173         protected function cacheURLsCallback($matches) {
 174                 $hash = '@@'.StringUtil::getHash(uniqid(microtime()).$matches[0]).'@@';
 175                 $this->cachedURLs[$hash] = $matches[0];
 176
 177                 return $hash;
 178         }
 179
 180         /**
 181          * Returns the hash for an matched e-mail in the message.
 182          *
 183          * @param       array           $matches
 184          * @return      string
 185          */
 186         protected function cacheEmailsCallback($matches) {
 187                 $hash = '@@'.StringUtil::getHash(uniqid(microtime()).$matches[0]).'@@';
 188                 $this->cachedEmails[$hash] = $matches[0];
 189
 190                 return $hash;
 191         }
 192
 193         /**
 194          * Reinserts cached URLs and e-mails.
 195          *
 196          * @param       string          $text
 197          * @return      string
 198          */
 199         protected function insertCachedURLs($text) {
 200                 foreach ($this->cachedURLs as $hash => $url) {
 201                         // add protocol if necessary
 202                         if (!preg_match("/[a-z]:\/\//si", $url)) {
 203                                 $url = 'http://'.$url;
 204                         }
 205
 206                         $text = str_replace($hash, StringUtil::getAnchorTag($url), $text);
 207                 }
 208
 209                 foreach ($this->cachedEmails as $hash => $email) {
 210                         $email = StringUtil::encodeHTML($email);
 211
 212                         $text = str_replace($hash, '<a href="mailto:'.$email.'">'.$email.'</a>', $text);
 213                 }
 214
 215                 return $text;
 216         }
 217
 218         /**
 219          * Parses smiley codes.
 220          *
 221          * @param       string          $text
 222          * @return      string          text
 223          */
 224         public function parseSmilies($text) {
 225                 foreach ($this->smilies as $code => $html) {
 226                         //$text = preg_replace('~(?<!&\w{2}|&\w{3}|&\w{4}|&\w{5}|&\w{6}|&#\d{2}|&#\d{3}|&#\d{4}|&#\d{5})'.preg_quote(StringUtil::encodeHTML($code), '~').'(?![^<]*>)~', $html, $text);
 227                         $text = preg_replace('~(?<=^|\s)'.preg_quote(StringUtil::encodeHTML($code), '~').'(?=$|\s|<br />)~', $html, $text);
 228                 }
 229
 230                 return $text;
 231         }
 232 }