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                 wfSuppressWarnings();
  39                 $this->mFP = fsockopen($this->mHost, $this->mPort, $errno, $errstr, 30);
  40                 wfRestoreWarnings();
  41                 if(!$this->mFP) {
  42                         return false;
  43                 }
  44                 return true;
  45         }
  46
  47         /**
  48          * Query the daemon and return the result
  49          *
  50          * @access private
  51          */
  52         function query($request) {
  53                 if(!$this->mConnected)
  54                         return false;
  55
  56                 fwrite($this->mFP, $request);
  57
  58                 $result=fgets($this->mFP, 1024);
  59
  60                 list($status, $len) = explode(" ", $result);
  61                 if($status == 'ERROR') {
  62                         //$len is actually the error code...
  63                         print "zhdaemon error $len<br />\n";
  64                         return false;
  65                 }
  66                 $bytesread=0;
  67                 $data='';
  68                 while(!feof($this->mFP) && $bytesread<$len) {
  69                         $str= fread($this->mFP, $len-$bytesread);
  70                         $bytesread += strlen($str);
  71                         $data .= $str;
  72                 }
  73                 //data should be of length $len. otherwise something is wrong
  74                 if(strlen($data) != $len)
  75                         return false;
  76                 return $data;
  77         }
  78
  79         /**
  80          * Convert the input to a different language variant
  81          *
  82          * @param string $text input text
  83          * @param string $tolang language variant
  84          * @return string the converted text
  85          * @access public
  86          */
  87         function convert($text, $tolang) {
  88                 $len = strlen($text);
  89                 $q = "CONV $tolang $len\n$text";
  90                 $result = $this->query($q);
  91                 if(!$result)
  92                         $result = $text;
  93                 return $result;
  94         }
  95
  96         /**
  97          * Convert the input to all possible variants
  98          *
  99          * @param string $text input text
 100          * @return array langcode => converted_string
 101          * @access public
 102          */
 103         function convertToAllVariants($text) {
 104                 $len = strlen($text);
 105                 $q = "CONV ALL $len\n$text";
 106                 $result = $this->query($q);
 107                 if(!$result)
 108                         return false;
 109                 list($infoline, $data) = explode('|', $result);
 110                 $info = explode(";", $infoline);
 111                 $ret = array();
 112                 $i=0;
 113                 foreach($info as $code => $len) {
 114                         $ret[strtolower($code)] = substr($data, $i, $len);
 115                         $i+=$len+1;
 116                 }
 117                 return $ret;
 118     }
 119         /**
 120          * Perform word segmentation
 121          *
 122          * @param string $text input text
 123          * @return string segmented text
 124          * @access public
 125          */
 126         function segment($text) {
 127                 $len = strlen($text);
 128                 $q = "SEG $len\n$text";
 129                 $result = $this->query($q);
 130                 if(!$result) {// fallback to character based segmentation
 131                         $result = ZhClientFake::segment($text);
 132                 }
 133                 return $result;
 134         }
 135
 136         /**
 137          * Close the connection
 138          *
 139          * @access public
 140          */
 141         function close() {
 142                 fclose($this->mFP);
 143         }
 144 }
 145
 146
 147 class ZhClientFake {
 148         function ZhClientFake() {
 149                 global $wgMemc, $wgDBname;
 150                 $this->zh2TW = $wgMemc->get($key1 = "$wgDBname:zhConvert:tw");
 151                 $this->zh2CN = $wgMemc->get($key2 = "$wgDBname:zhConvert:cn");
 152                 $this->zh2SG = $wgMemc->get($key3 = "$wgDBname:zhConvert:sg");
 153                 $this->zh2HK = $wgMemc->get($key4 = "$wgDBname:zhConvert:hk");
 154                 if(empty($this->zh2TW) || empty($this->zh2CN) || empty($this->zh2SG) || empty($this->zh2HK)) {
 155                         require("includes/ZhConversion.php");
 156                         $this->zh2TW = $zh2TW;
 157                         $this->zh2CN = $zh2CN;
 158                         $this->zh2HK = $zh2HK;
 159                         $this->zh2SG = $zh2SG;
 160                         $wgMemc->set($key1, $this->zh2TW);
 161                         $wgMemc->set($key2, $this->zh2CN);
 162                         $wgMemc->set($key3, $this->zh2SG);
 163                         $wgMemc->set($key4, $this->zh2HK);
 164                 }
 165         }
 166
 167         function isconnected() {
 168                 return true;
 169         }
 170
 171         /**
 172          * Convert to zh-tw
 173          *
 174          * @access private
 175          */
 176         function zh2tw($text) {
 177                 return strtr($text, $this->zh2TW);
 178         }
 179
 180         /**
 181          * Convert to zh-cn
 182          *
 183          * @access private
 184          */
 185         function zh2cn($text) {
 186                 return strtr($text, $this->zh2CN);
 187         }
 188
 189         /**
 190          * Convert to zh-sg
 191          *
 192          * @access private
 193          */
 194         function zh2sg($text) {
 195                 return strtr(strtr($text, $this->zh2CN), $this->zh2SG);
 196         }
 197
 198         /**
 199          * Convert to zh-hk
 200          *
 201          * @access private
 202          */
 203         function zh2hk($text) {
 204                 return strtr(strtr($text, $this->zh2TW), $this->zh2HK);
 205         }
 206
 207         /**
 208          * Convert the input to a different language variant
 209          *
 210          * @param string $text input text
 211          * @param string $tolang language variant
 212          * @return string the converted text
 213          * @access public
 214          */
 215         function convert($text, $tolang) {
 216                 $t = '';
 217                 switch($tolang) {
 218         case 'zh-cn':
 219                         $t = $this->zh2cn($text);
 220                         break;
 221                 case 'zh-tw':
 222                         $t = $this->zh2tw($text);
 223                         break;
 224                 case 'zh-sg':
 225                         $t = $this->zh2sg($text);
 226                         break;
 227                 case 'zh-hk':
 228                         $t = $this->zh2hk($text);
 229                         break;
 230                 default:
 231                         $t = $text;
 232                 }
 233                 return $t;
 234         }
 235
 236         function convertToAllVariants($text) {
 237                 $ret = array();
 238                 $ret['zh-cn'] = $this->zh2cn($text);
 239                 $ret['zh-tw'] = $this->zh2tw($text);
 240                 $ret['zh-sg'] = $this->zh2sg($text);
 241                 $ret['zh-hk'] = $this->zh2hk($text);
 242                 return $ret;
 243         }
 244
 245         /**
 246          * Perform "fake" word segmentation, i.e. treating each character as a word
 247          *
 248          * @param string $text input text
 249          * @return string segmented text
 250          * @access public
 251          */
 252         function segment($text) {
 253                 /* adapted from LanguageZh_cn.stripForSearch()
 254                         here we will first separate the single characters,
 255                         and let the caller conver it to hex
 256         */
 257                 if( function_exists( 'mb_strtolower' ) ) {
 258                         return preg_replace(
 259                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 260                                 "' ' .\"$1\"",
 261                                 mb_strtolower( $text ) );
 262                 } else {
 263                         global $wikiLowerChars;
 264                         return preg_replace(
 265                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 266                                 "' ' . strtr( \"\$1\", \$wikiLowerChars )",
 267                                 $text );
 268                 }
 269         }
 270
 271         /**
 272          * Close the fake connection
 273          *
 274          * @access public
 275          */
 276         function close() {      }
 277 }
 278
 279 ?>