includes/ZhClient.php

   1 <?php
   2 /**
   3  * Client for querying zhdaemon
   4  *
   5  * @package MediaWiki
   6  */
   7
   8 class ZhClient {
   9         var $mHost, $mPort, $mFP, $mConnected;
  10
  11         /**
  12          * Constructor
  13          *
  14          * @access private
  15          */
  16         function ZhClient($host, $port) {
  17                 $this->mHost = $host;
  18                 $this->mPort = $port;
  19                 $this->mConnected = $this->connect();
  20         }
  21
  22         /**
  23          * Check if connection to zhdaemon is successful
  24          *
  25          * @access public
  26          */
  27         function isconnected() {
  28                 return $this->mConnected;
  29         }
  30
  31         /**
  32          * Establish conncetion
  33          *
  34          * @access private
  35          */
  36         function connect() {
  37                 wfSuppressWarnings();
  38                 $this->mFP = fsockopen($this->mHost, $this->mPort, $errno, $errstr, 30);
  39                 wfRestoreWarnings();
  40                 if(!$this->mFP) {
  41                         return false;
  42                 }
  43                 return true;
  44         }
  45
  46         /**
  47          * Query the daemon and return the result
  48          *
  49          * @access private
  50          */
  51         function query($request) {
  52                 if(!$this->mConnected)
  53                         return false;
  54
  55                 fwrite($this->mFP, $request);
  56
  57                 $result=fgets($this->mFP, 1024);
  58
  59                 list($status, $len) = explode(" ", $result);
  60                 if($status == 'ERROR') {
  61                         //$len is actually the error code...
  62                         print "zhdaemon error $len<br />\n";
  63                         return false;
  64                 }
  65                 $bytesread=0;
  66                 $data='';
  67                 while(!feof($this->mFP) && $bytesread<$len) {
  68                         $str= fread($this->mFP, $len-$bytesread);
  69                         $bytesread += strlen($str);
  70                         $data .= $str;
  71                 }
  72                 //data should be of length $len. otherwise something is wrong
  73                 if(strlen($data) != $len)
  74                         return false;
  75                 return $data;
  76         }
  77
  78         /**
  79          * Convert the input to a different language variant
  80          *
  81          * @param string $text input text
  82          * @param string $tolang language variant
  83          * @return string the converted text
  84          * @access public
  85          */
  86         function convert($text, $tolang) {
  87                 $len = strlen($text);
  88                 $q = "CONV $tolang $len\n$text";
  89                 $result = $this->query($q);
  90                 if(!$result)
  91                         $result = $text;
  92                 return $result;
  93         }
  94
  95         /**
  96          * Convert the input to all possible variants
  97          *
  98          * @param string $text input text
  99          * @return array langcode => converted_string
 100          * @access public
 101          */
 102         function convertToAllVariants($text) {
 103                 $len = strlen($text);
 104                 $q = "CONV ALL $len\n$text";
 105                 $result = $this->query($q);
 106                 if(!$result)
 107                         return false;
 108                 list($infoline, $data) = explode('|', $result, 2);
 109                 $info = explode(";", $infoline);
 110                 $ret = array();
 111                 $i=0;
 112                 foreach($info as $variant) {
 113                         list($code, $len) = explode(' ', $variant);
 114                         $ret[strtolower($code)] = substr($data, $i, $len);
 115                         $r = $ret[strtolower($code)];
 116                         $i+=$len;
 117                 }
 118                 return $ret;
 119     }
 120         /**
 121          * Perform word segmentation
 122          *
 123          * @param string $text input text
 124          * @return string segmented text
 125          * @access public
 126          */
 127         function segment($text) {
 128                 $len = strlen($text);
 129                 $q = "SEG $len\n$text";
 130                 $result = $this->query($q);
 131                 if(!$result) {// fallback to character based segmentation
 132                         $result = ZhClientFake::segment($text);
 133                 }
 134                 return $result;
 135         }
 136
 137         /**
 138          * Close the connection
 139          *
 140          * @access public
 141          */
 142         function close() {
 143                 fclose($this->mFP);
 144         }
 145 }
 146
 147
 148 class ZhClientFake {
 149         function ZhClientFake() {
 150                 global $wgMemc, $wgDBname;
 151                 $this->mZh2TW = $wgMemc->get($key1 = "$wgDBname:zhConvert:tw");
 152                 $this->mZh2CN = $wgMemc->get($key2 = "$wgDBname:zhConvert:cn");
 153                 $this->mZh2SG = $wgMemc->get($key3 = "$wgDBname:zhConvert:sg");
 154                 $this->mZh2HK = $wgMemc->get($key4 = "$wgDBname:zhConvert:hk");
 155                 if(empty($this->mZh2TW) || empty($this->mZh2CN) || empty($this->mZh2SG) || empty($this->mZh2HK)) {
 156                         require("includes/ZhConversion.php");
 157                         $this->mZh2TW = $zh2TW;
 158                         $this->mZh2CN = $zh2CN;
 159                         $this->mZh2HK = $zh2HK;
 160                         $this->mZh2SG = $zh2SG;
 161                         $wgMemc->set($key1, $this->mZh2TW);
 162                         $wgMemc->set($key2, $this->mZh2CN);
 163                         $wgMemc->set($key3, $this->mZh2SG);
 164                         $wgMemc->set($key4, $this->mZh2HK);
 165                 }
 166         }
 167
 168         function isconnected() {
 169                 return true;
 170         }
 171
 172         /**
 173          * Convert to zh-tw
 174          *
 175          * @access private
 176          */
 177         function zh2tw($text) {
 178                 return strtr($text, $this->mZh2TW);
 179         }
 180
 181         /**
 182          * Convert to zh-cn
 183          *
 184          * @access private
 185          */
 186         function zh2cn($text) {
 187                 return strtr($text, $this->mZh2CN);
 188         }
 189
 190         /**
 191          * Convert to zh-sg
 192          *
 193          * @access private
 194          */
 195         function zh2sg($text) {
 196                 return strtr(strtr($text, $this->mZh2CN), $this->mZh2SG);
 197         }
 198
 199         /**
 200          * Convert to zh-hk
 201          *
 202          * @access private
 203          */
 204         function zh2hk($text) {
 205                 return strtr(strtr($text, $this->mZh2TW), $this->mZh2HK);
 206         }
 207
 208         /**
 209          * Convert the input to a different language variant
 210          *
 211          * @param string $text input text
 212          * @param string $tolang language variant
 213          * @return string the converted text
 214          * @access public
 215          */
 216         function convert($text, $tolang) {
 217                 $t = '';
 218                 switch($tolang) {
 219         case 'zh-cn':
 220                         $t = $this->zh2cn($text);
 221                         break;
 222                 case 'zh-tw':
 223                         $t = $this->zh2tw($text);
 224                         break;
 225                 case 'zh-sg':
 226                         $t = $this->zh2sg($text);
 227                         break;
 228                 case 'zh-hk':
 229                         $t = $this->zh2hk($text);
 230                         break;
 231                 default:
 232                         $t = $text;
 233                 }
 234                 return $t;
 235         }
 236
 237         function convertToAllVariants($text) {
 238                 $ret = array();
 239                 $ret['zh-cn'] = $this->zh2cn($text);
 240                 $ret['zh-tw'] = $this->zh2tw($text);
 241                 $ret['zh-sg'] = $this->zh2sg($text);
 242                 $ret['zh-hk'] = $this->zh2hk($text);
 243                 return $ret;
 244         }
 245
 246         /**
 247          * Perform "fake" word segmentation, i.e. treating each character as a word
 248          *
 249          * @param string $text input text
 250          * @return string segmented text
 251          * @access public
 252          */
 253         function segment($text) {
 254                 /* adapted from LanguageZh_cn::stripForSearch()
 255                         here we will first separate the single characters,
 256                         and let the caller conver it to hex
 257         */
 258                 if( function_exists( 'mb_strtolower' ) ) {
 259                         return preg_replace(
 260                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 261                                 "' ' .\"$1\"",
 262                                 mb_strtolower( $text ) );
 263                 } else {
 264                         global $wikiLowerChars;
 265                         return preg_replace(
 266                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 267                                 "' ' . strtr( \"\$1\", \$wikiLowerChars )",
 268                                 $text );
 269                 }
 270         }
 271
 272         /**
 273          * Close the fake connection
 274          *
 275          * @access public
 276          */
 277         function close() {      }
 278 }
 279
 280 ?>