includes/ZhClient.php

   1 <?php
   2 /**
   3  * Client for querying zhdaemon
   4  *
   5  * @package MediaWiki
   6  * @version $Id$
   7  */
   8
   9 class ZhClient {
  10         var $mHost, $mPort, $mFP, $mConnected;
  11
  12         /**
  13          * Constructor
  14          *
  15          * @access private
  16          */
  17         function ZhClient($host, $port) {
  18                 $this->mHost = $host;
  19                 $this->mPort = $port;
  20                 $this->mConnected = $this->connect();
  21         }
  22
  23         /**
  24          * Check if connection to zhdaemon is successful
  25          *
  26          * @access public
  27          */
  28         function isconnected() {
  29                 return $this->mConnected;
  30         }
  31
  32         /**
  33          * Establish conncetion
  34          *
  35          * @access private
  36          */
  37         function connect() {
  38                 $this->mFP = fsockopen($this->mHost, $this->mPort, $errno, $errstr, 30);
  39                 if(!$this->mFP) {
  40                         return false;
  41                 }
  42                 return true;
  43         }
  44
  45         /**
  46          * Query the daemon and return the result
  47          *
  48          * @access private
  49          */
  50         function query($request) {
  51                 if(!$this->mConnected)
  52                         return false;
  53
  54                 fwrite($this->mFP, $request);
  55
  56                 $result=fgets($this->mFP, 1024);
  57
  58                 list($status, $len) = explode(" ", $result);
  59                 if($status == 'ERROR') {
  60                         //$len is actually the error code...
  61                         print "zhdaemon error $len<br />\n";
  62                         return false;
  63                 }
  64                 $bytesread=0;
  65                 $data='';
  66                 while(!feof($this->mFP) && $bytesread<$len) {
  67                         $str= fread($this->mFP, $len-$bytesread);
  68                         $bytesread += strlen($str);
  69                         $data .= $str;
  70                 }
  71                 return $data;
  72         }
  73
  74         /**
  75          * Convert the input to a different language variant
  76          *
  77          * @param string $text input text
  78          * @param string $tolang language variant
  79          * @return string the converted text
  80          * @access public
  81          */
  82         function convert($text, $tolang) {
  83                 $len = strlen($text);
  84                 $q = "CONV $tolang $len\n$text";
  85                 $result = $this->query($q);
  86                 if(!$result)
  87                         $result = $text;
  88                 return $result;
  89         }
  90
  91         /**
  92          * Perform word segmentation
  93          *
  94          * @param string $text input text
  95          * @return string segmented text
  96          * @access public
  97          */
  98         function segment($text) {
  99                 $len = strlen($text);
 100                 $q = "SEG $len\n$text";
 101                 $result = $this->query($q);
 102                 if(!$result)
 103                         $result = $text;
 104                 return $result;
 105         }
 106
 107         /**
 108          * Close the connection
 109          *
 110          * @access public
 111          */
 112         function close() {
 113                 fclose($this->mFP);
 114         }
 115 }
 116
 117
 118 class ZhClientFake {
 119
 120         function ZhClientFake() {
 121                 global $wgMemc, $wgDBname;
 122                 $this->zh2TW = $wgMemc->get($key1 = "$wgDBname:zhConvert:tw");
 123                 $this->zh2CN = $wgMemc->get($key2 = "$wgDBname:zhConvert:cn");
 124                 $this->zh2SG = $wgMemc->get($key3 = "$wgDBname:zhConvert:sg");
 125                 $this->zh2HK = $wgMemc->get($key4 = "$wgDBname:zhConvert:hk");
 126                 if(empty($this->zh2TW) || empty($this->zh2CN) || empty($this->zh2SG) || empty($this->zh2HK)) {
 127                         require_once("includes/ZhConversion.php");
 128                         $this->zh2TW = $zh2TW;
 129                         $this->zh2CN = $zh2CN;
 130                         $this->zh2HK = $zh2HK;
 131                         $this->zh2SG = $zh2SG;
 132                         $wgMemc->set($key1, $this->zh2TW);
 133                         $wgMemc->set($key2, $this->zh2CN);
 134                         $wgMemc->set($key3, $this->zh2SG);
 135                         $wgMemc->set($key4, $this->zh2HK);
 136                 }
 137         }
 138
 139         function isconnected() {
 140                 return true;
 141         }
 142
 143         /**
 144          * Convert to zh-tw
 145          *
 146          * @access private
 147          */
 148         function zh2tw($text) {
 149                 return strtr($text, $this->zh2TW);
 150         }
 151
 152         /**
 153          * Convert to zh-cn
 154          *
 155          * @access private
 156          */
 157         function zh2cn($text) {
 158                 return strtr($text, $this->zh2CN);
 159         }
 160
 161         /**
 162          * Convert to zh-sg
 163          *
 164          * @access private
 165          */
 166         function zh2sg($text) {
 167                 return strtr(strtr($text, $this->zh2CN), $this->zh2SG);
 168         }
 169
 170         /**
 171          * Convert to zh-hk
 172          *
 173          * @access private
 174          */
 175         function zh2hk($text) {
 176                 return strtr(strtr($text, $this->zh2TW), $this->zh2HK);
 177         }
 178
 179         /**
 180          * Convert the input to a different language variant
 181          *
 182          * @param string $text input text
 183          * @param string $tolang language variant
 184          * @return string the converted text
 185          * @access public
 186          */
 187         function convert($text, $tolang) {
 188                 $t = '';
 189                 switch($tolang) {
 190         case 'zh-cn':
 191                         $t = $this->zh2cn($text);
 192                         break;
 193                 case 'zh-tw':
 194                         $t = $this->zh2tw($text);
 195                         break;
 196                 case 'zh-sg':
 197                         $t = $this->zh2sg($text);
 198                         break;
 199                 case 'zh-hk':
 200                         $t = $this->zh2hk($text);
 201                         break;
 202                 default:
 203                         $t = $text;
 204                 }
 205                 return $t;
 206         }
 207
 208         /**
 209          * Perform "fake" word segmentation, i.e. treating each character as a word
 210          *
 211          * @param string $text input text
 212          * @return string segmented text
 213          * @access public
 214          */
 215         function segment($text) {
 216                 /* copied from LanguageZh_cn.stripForSearch() */
 217                 if( function_exists( 'mb_strtolower' ) ) {
 218                         return preg_replace(
 219                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 220                                 "' U8' . bin2hex( \"$1\" )",
 221                                 mb_strtolower( $string ) );
 222                 } else {
 223                         global $wikiLowerChars;
 224                         return preg_replace(
 225                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 226                                 "' U8' . bin2hex( strtr( \"\$1\", \$wikiLowerChars ) )",
 227                                 $string );
 228                 }
 229         }
 230
 231         /**
 232          * Close the fake connection
 233          *
 234          * @access public
 235          */
 236         function close() {      }
 237 }
 238
 239 ?>