DedeBIZ
/
DedeV6
Watch 2
Star 0
Fork 0
Code Pull Requests 0 Releases 25 Activity
国内流行的内容管理系统（CMS）多端全媒体解决方案 https://www.dedebiz.com
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
275 Commits
3 Branches
110MB
Tree: 66f402eac8
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '66f402eac8'
${ noResults }
DedeV6/src/include/splitword.class.php

1134 lines
38KB
Raw Blame History 

  1. <?php

  2. /**

  3.  * Unicode编码词典的php分词器

  4.  *

  5.  *  1、只适用于php5，必要函数 iconv

  6.  *  2、本程序是使用RMM逆向匹配算法进行分词的，词库需要特别编译，本类里提供了 MakeDict() 方法

  7.  *  3、简单操作流程： SetSource -> StartAnalysis -> Get***Result

  8.  *  4、对主词典使用特殊格式进行编码, 不需要载入词典到内存操作

  9.  *

  10.  * @version        $Id: splitword.class.php 2 11:45 2011-2-14 itplato $

  11.  * @package        DedeBIZ.Libraries

  12.  * @copyright      Copyright (c) 2020, DedeBIZ.COM

  13.  * @license        https://www.dedebiz.com/license

  14.  * @link           https://www.dedebiz.com

  15.  */

  16. 

  17. //常量定义

  18. define('_SP_', chr(0xFF).chr(0xFE)); 

  19. define('UCS2', 'ucs-2be');

  20. class SplitWord

  21. {

  22.     

  23.     //hash算法选项

  24.     var $mask_value = 0xFFFF;

  25.     

  26.     //输入和输出的字符编码（只允许 utf-8、gbk/gb2312/gb18030、big5 三种类型）  

  27.     var $sourceCharSet = 'utf-8';

  28.     var $targetCharSet = 'utf-8';

  29.     

  30.     //生成的分词结果数据类型 1 为全部， 2为 词典词汇及单个中日韩简繁字符及英文， 3 为词典词汇及英文

  31.     var $resultType = 1;

  32.     

  33.     //句子长度小于这个数值时不拆分，notSplitLen = n(个汉字) * 2 + 1

  34.     var $notSplitLen = 5;

  35.     

  36.     //把英文单词全部转小写

  37.     var $toLower = FALSE;

  38.     

  39.     //使用最大切分模式对二元词进行消岐

  40.     var $differMax = FALSE;

  41.     

  42.     //尝试合并单字

  43.     var $unitWord = TRUE;

  44.     

  45.     //初始化类时直接加载词典

  46.     var $loadInit = TRUE;

  47.     

  48.     //使用热门词优先模式进行消岐

  49.     var $differFreq = FALSE;

  50.     

  51.     //被转换为unicode的源字符串

  52.     var $sourceString = '';

  53.     

  54.     //附加词典

  55.     var $addonDic = array();

  56.     var $addonDicFile = 'data/words_addons.dic';

  57.     

  58.     //主词典 

  59.     var $dicStr = '';

  60.     var $mainDic = array();

  61.     var $mainDicHand = FALSE;

  62.     var $mainDicInfos = array();

  63.     var $mainDicFile = 'data/base_dic_full.dic';

  64.     //是否直接载入词典（选是载入速度较慢，但解析较快；选否载入较快，但解析较慢，需要时才会载入特定的词条）

  65.     var $mainDicFileZip = 'data/base_dic_full.zip';

  66.     var $isLoadAll = FALSE;

  67.     var $isUnpacked = FALSE;

  68.     

  69.     //主词典词语最大长度 x / 2

  70.     var $dicWordMax = 14;

  71.     //粗分后的数组（通常是截取句子等用途）

  72.     var $simpleResult = array();

  73.     //最终结果(用空格分开的词汇列表)

  74.     var $finallyResult = '';

  75.     

  76.     //是否已经载入词典

  77.     var $isLoadDic = FALSE;

  78.     //系统识别或合并的新词

  79.     var $newWords = array();

  80.     var $foundWordStr = '';

  81.     //词库载入时间

  82.     var $loadTime = 0;

  83.     

  84.     /**

  85.      * 构造函数

  86.      * @param $source_charset

  87.      * @param $target_charset

  88.      * @param $load_alldic 

  89.      * @param $source

  90.      *

  91.      * @return void

  92.      */

  93.     function __construct($source_charset='utf-8', $target_charset='utf-8', $load_all=TRUE, $source='')

  94.     {

  95.         $this->SetSource( $source, $source_charset, $target_charset );

  96.         $this->isLoadAll = $load_all;

  97.         if(file_exists(DEDEINC.'/'.$this->mainDicFile)) $this->isUnpacked = TRUE;

  98.         if($this->loadInit) $this->LoadDict();

  99.     }

  100.     

  101.     function SplitWord($source_charset='utf-8', $target_charset='utf-8', $load_all=TRUE, $source='')

  102.     {

  103.         $this->__construct($source_charset, $target_charset, $load_all, $source);

  104.     }

  105.     

  106.    /**

  107.     * 析构函数

  108.     */

  109.     function __destruct()

  110.     {

  111.         if( $this->mainDicHand !== FALSE )

  112.         {

  113.             @fclose( $this->mainDicHand );

  114.         }

  115.     }

  116.     

  117.     /**

  118.      * 根据字符串计算key索引

  119.      * @param $key

  120.      * @return short int

  121.      */

  122.     function _get_index( $key )

  123.     {

  124.         $l = strlen($key);

  125.         $h = 0x238f13af;

  126.         while ($l--)

  127.         {

  128.             $h += ($h << 5);

  129.             $h ^= ord($key[$l]);

  130.             $h &= 0x7fffffff;

  131.         }

  132.         return ($h % $this->mask_value);

  133.     }

  134.     

  135.     /**

  136.      * 从文件获得词

  137.      * @param $key

  138.      * @param $type (类型 word 或 key_groups)

  139.      * @return short int

  140.      */

  141.     function GetWordInfos( $key, $type='word' )

  142.     {

  143.         if( !$this->mainDicHand )

  144.         {

  145.             $this->mainDicHand = fopen($this->mainDicFile, 'r');

  146.         }

  147.         $p = 0;

  148.         $keynum = $this->_get_index( $key );

  149.         if( isset($this->mainDicInfos[ $keynum ]) )

  150.         {

  151.             $data = $this->mainDicInfos[ $keynum ];

  152.         }

  153.         else

  154.         {

  155.             //rewind( $this->mainDicHand );

  156.             $move_pos = $keynum * 8;

  157.             fseek($this->mainDicHand, $move_pos, SEEK_SET);

  158.             $dat = fread($this->mainDicHand, 8);

  159.             $arr = unpack('I1s/n1l/n1c', $dat);

  160.             if( $arr['l'] == 0 )

  161.             {

  162.                 return FALSE;

  163.             }

  164.             fseek($this->mainDicHand, $arr['s'], SEEK_SET);

  165.             $data = @unserialize(fread($this->mainDicHand, $arr['l']));

  166.             $this->mainDicInfos[ $keynum ] = $data;

  167.        }

  168.        if( !is_array($data) || !isset($data[$key]) ) 

  169.        {

  170.            return FALSE;

  171.        }

  172.        return ($type=='word' ? $data[$key] : $data);

  173.     }

  174.     

  175.     /**

  176.      * 设置源字符串

  177.      * @param $source

  178.      * @param $source_charset

  179.      * @param $target_charset

  180.      *

  181.      * @return bool

  182.      */

  183.     function SetSource( $source, $source_charset='utf-8', $target_charset='utf-8' )

  184.     {

  185.         $this->sourceCharSet = strtolower($source_charset);

  186.         $this->targetCharSet = strtolower($target_charset);

  187.         $this->simpleResult = array();

  188.         $this->finallyResult = array();

  189.         $this->finallyIndex = array();

  190.         if( $source != '' )

  191.         {

  192.             $rs = TRUE;

  193.             if( preg_match("/^utf/", $source_charset) ) {

  194.                 $this->sourceString = @iconv('utf-8', UCS2, $source);

  195.             }

  196.             else if( preg_match("/^gb/", $source_charset) ) {

  197.                 $this->sourceString = @iconv('utf-8', UCS2, iconv('gb18030', 'utf-8', $source));

  198.             }

  199.             else if( preg_match("/^big/", $source_charset) ) {

  200.                 $this->sourceString = @iconv('utf-8', UCS2, iconv('big5', 'utf-8', $source));

  201.             }

  202.             else {

  203.                 $rs = FALSE;

  204.             }

  205.         }

  206.         else

  207.         {

  208.            $rs = FALSE;

  209.         }

  210.         return $rs;

  211.     }

  212.     

  213.     /**

  214.      * 设置结果类型(只在获取finallyResult才有效)

  215.      * @param $rstype 1 为全部， 2去除特殊符号

  216.      *

  217.      * @return void

  218.      */

  219.     function SetResultType( $rstype )

  220.     {

  221.         $this->resultType = $rstype;

  222.     }

  223.     

  224.     /**

  225.      * 载入词典

  226.      *

  227.      * @return void

  228.      */

  229.     function LoadDict( $maindic='' )

  230.     {

  231. 		$this->addonDicFile = DEDEINC.'/'.$this->addonDicFile;

  232. 		$this->mainDicFile = DEDEINC.'/'.$this->mainDicFile;

  233. 		$this->mainDicFileZip = DEDEINC.'/'.$this->mainDicFileZip;

  234.         $startt = microtime(TRUE);

  235.         //正常读取文件

  236.         $dicAddon = $this->addonDicFile;

  237.         if($maindic=='' || !file_exists($maindic) )

  238.         {

  239.             $dicWords = $this->mainDicFile ;

  240.         }

  241.         else

  242.         {

  243.             $dicWords = $maindic;

  244.             $this->mainDicFile = $maindic;

  245.         }

  246.         

  247.         //加载主词典（只打开）

  248.         if($this->isUnpacked){

  249.         	$this->mainDicHand = fopen($dicWords, 'r');

  250.         }else{

  251.         	$this->InportDict($this->mainDicFileZip);

  252.         }

  253.         

  254.         //载入副词典

  255.         $hw = '';

  256.         $ds = file($dicAddon);

  257.         foreach($ds as $d)

  258.         {

  259.             $d = trim($d);

  260.             if($d=='') continue;

  261.             $estr = substr($d, 1, 1);

  262.             if( $estr==':' ) {

  263.                 $hw = substr($d, 0, 1);

  264.             }

  265.             else

  266.             {

  267.                 $spstr = _SP_;

  268.                 $spstr = iconv(UCS2, 'utf-8', $spstr);

  269.                 $ws = explode(',', $d);

  270.                 $wall = iconv('utf-8', UCS2, join($spstr, $ws));

  271.                 $ws = explode(_SP_, $wall);

  272.                 foreach($ws as $estr)

  273.                 {

  274.                     $this->addonDic[$hw][$estr] = strlen($estr);

  275.                 }

  276.             }

  277.         }

  278.         $this->loadTime = microtime(TRUE) - $startt;

  279.         $this->isLoadDic = TRUE;

  280.     }

  281.     

  282.    /**

  283.     * 检测某个词是否存在

  284.     */

  285.     function IsWord( $word )

  286.     {

  287.          $winfos = $this->GetWordInfos( $word );

  288.          return ($winfos !== FALSE);

  289.     }

  290.     

  291.     /**

  292.      * 获得某个词的词性及词频信息

  293.      * @parem $word unicode编码的词

  294.      * @return void

  295.      */

  296.      function GetWordProperty($word)

  297.      {

  298.         if( strlen($word)<4 )

  299.         {

  300.             return '/s';

  301.         }

  302.         $infos = $this->GetWordInfos($word);

  303.         return isset($infos[1]) ? "/{$infos[1]}{$infos[0]}" : "/s";

  304.      }

  305.     

  306.     /**

  307.      * 指定某词的词性信息（通常是新词）

  308.      * @parem $word unicode编码的词

  309.      * @parem $infos array('c' => 词频, 'm' => 词性);

  310.      * @return void;

  311.      */

  312.     function SetWordInfos($word, $infos)

  313.     {

  314.         if( strlen($word)<4 )

  315.         {

  316.             return ;

  317.         }

  318.         if( isset($this->mainDicInfos[$word]) )

  319.         {

  320.             $this->newWords[$word]++;

  321.             $this->mainDicInfos[$word]['c']++;

  322.         }

  323.         else

  324.         {

  325.             $this->newWords[$word] = 1;

  326.             $this->mainDicInfos[$word] = $infos;

  327.         }

  328.     }

  329.     

  330.     /**

  331.      * 开始执行分析

  332.      * @parem bool optimize 是否对结果进行优化

  333.      * @return bool

  334.      */

  335.     function StartAnalysis($optimize=TRUE)

  336.     {

  337.         if( !$this->isLoadDic )

  338.         {

  339.             $this->LoadDict();

  340.         }

  341.         $this->simpleResult = $this->finallyResult = array();

  342.         $this->sourceString .= chr(0).chr(32);

  343.         $slen = strlen($this->sourceString);

  344.         $sbcArr = array();

  345.         $j = 0;

  346.         //全角与半角字符对照表

  347.         for($i=0xFF00; $i < 0xFF5F; $i++)

  348.         {

  349.             $scb = 0x20 + $j;

  350.             $j++;

  351.             $sbcArr[$i] = $scb;

  352.         }

  353.         //对字符串进行粗分

  354.         $onstr = '';

  355.         $lastc = 1; //1 中/韩/日文, 2 英文/数字/符号('.', '@', '#', '+'), 3 ANSI符号 4 纯数字 5 非ANSI符号或不支持字符

  356.         $s = 0;

  357.         $ansiWordMatch = "[0-9a-z@#%\+\.-]";

  358.         $notNumberMatch = "[a-z@#%\+]";

  359.         for($i=0; $i < $slen; $i++)

  360.         {

  361.             $c = $this->sourceString[$i].$this->sourceString[++$i];

  362.             $cn = hexdec(bin2hex($c));

  363.             $cn = isset($sbcArr[$cn]) ? $sbcArr[$cn] : $cn;

  364.             //ANSI字符

  365.             if($cn < 0x80)

  366.             {

  367.                 if( preg_match('/'.$ansiWordMatch.'/i', chr($cn)) )

  368.                 {

  369.                     if( $lastc != 2 && $onstr != '') {

  370.                         $this->simpleResult[$s]['w'] = $onstr;

  371.                         $this->simpleResult[$s]['t'] = $lastc;

  372.                         $this->_deep_analysis($onstr, $lastc, $s, $optimize);

  373.                         $s++;

  374.                         $onstr = '';

  375.                     }

  376.                     $lastc = 2;

  377.                     $onstr .= chr(0).chr($cn);

  378.                 }

  379.                 else

  380.                 {

  381.                     if( $onstr != '' )

  382.                     {

  383.                         $this->simpleResult[$s]['w'] = $onstr;

  384.                         if( $lastc==2 )

  385.                         {

  386.                             if( !preg_match('/'.$notNumberMatch.'/i', iconv(UCS2, 'utf-8', $onstr)) ) $lastc = 4;

  387.                         }

  388.                         $this->simpleResult[$s]['t'] = $lastc;

  389.                         if( $lastc != 4 ) $this->_deep_analysis($onstr, $lastc, $s, $optimize);

  390.                         $s++;

  391.                     }

  392.                     $onstr = '';

  393.                     $lastc = 3;

  394.                     if($cn < 31)

  395.                     {

  396.                         continue;

  397.                     }

  398.                     else

  399.                     {

  400.                         $this->simpleResult[$s]['w'] = chr(0).chr($cn);

  401.                         $this->simpleResult[$s]['t'] = 3;

  402.                         $s++;

  403.                     }

  404.                 }

  405.             }

  406.             //普通字符

  407.             else

  408.             {

  409.                 //正常文字

  410.                 if( ($cn>0x3FFF && $cn < 0x9FA6) || ($cn>0xF8FF && $cn < 0xFA2D)

  411.                     || ($cn>0xABFF && $cn < 0xD7A4) || ($cn>0x3040 && $cn < 0x312B) )

  412.                 {

  413.                     if( $lastc != 1 && $onstr != '')

  414.                     {

  415.                         $this->simpleResult[$s]['w'] = $onstr;

  416.                         if( $lastc==2 )

  417.                         {

  418.                             if( !preg_match('/'.$notNumberMatch.'/i', iconv(UCS2, 'utf-8', $onstr)) ) $lastc = 4;

  419.                         }

  420.                         $this->simpleResult[$s]['t'] = $lastc;

  421.                         if( $lastc != 4 ) $this->_deep_analysis($onstr, $lastc, $s, $optimize);

  422.                         $s++;

  423.                         $onstr = '';

  424.                     }

  425.                     $lastc = 1;

  426.                     $onstr .= $c;

  427.                 }

  428.                 //特殊符号

  429.                 else

  430.                 {

  431.                     if( $onstr != '' )

  432.                     {

  433.                         $this->simpleResult[$s]['w'] = $onstr;

  434.                         if( $lastc==2 )

  435.                         {

  436.                             if( !preg_match('/'.$notNumberMatch.'/i', iconv(UCS2, 'utf-8', $onstr)) ) $lastc = 4;

  437.                         }

  438.                         $this->simpleResult[$s]['t'] = $lastc;

  439.                         if( $lastc != 4 ) $this->_deep_analysis($onstr, $lastc, $s, $optimize);

  440.                         $s++;

  441.                     }

  442.                     

  443.                     //检测书名

  444.                     if( $cn == 0x300A )

  445.                     {

  446.                         $tmpw = '';

  447.                         $n = 1;

  448.                         $isok = FALSE;

  449.                         $ew = chr(0x30).chr(0x0B);

  450.                         while(TRUE)

  451.                         {

  452. 							if(!isset($this->sourceString[$i+$n]) && !isset($this->sourceString[$i+$n+1]))

  453. 							break;

  454.                             $w = $this->sourceString[$i+$n].$this->sourceString[$i+$n+1];

  455.                             if( $w == $ew )

  456.                             {

  457.                                 $this->simpleResult[$s]['w'] = $c;

  458.                                 $this->simpleResult[$s]['t'] = 5;

  459.                                 $s++;

  460.                         

  461.                                 $this->simpleResult[$s]['w'] = $tmpw;

  462.                                 $this->newWords[$tmpw] = 1;

  463.                                 if( !isset($this->newWords[$tmpw]) )

  464.                                 {

  465.                                     $this->foundWordStr .= $this->_out_string_encoding($tmpw).'/nb, ';

  466.                                     $this->SetWordInfos($tmpw, array('c'=>1, 'm'=>'nb'));

  467.                                 }

  468.                                 $this->simpleResult[$s]['t'] = 13;

  469.                                 

  470.                                 $s++;

  471. 

  472.                                 //最大切分模式对书名继续分词

  473.                                 if( $this->differMax )

  474.                                 {

  475.                                     $this->simpleResult[$s]['w'] = $tmpw;

  476.                                     $this->simpleResult[$s]['t'] = 21;

  477.                                     $this->_deep_analysis($tmpw, $lastc, $s, $optimize);

  478.                                     $s++;

  479.                                 }

  480.                                 

  481.                                 $this->simpleResult[$s]['w'] = $ew;

  482.                                 $this->simpleResult[$s]['t'] =  5;

  483.                                 $s++;

  484.                         

  485.                                 $i = $i + $n + 1;

  486.                                 $isok = TRUE;

  487.                                 $onstr = '';

  488.                                 $lastc = 5;

  489.                                 break;

  490.                             }

  491.                             else

  492.                             {

  493.                                 $n = $n+2;

  494.                                 $tmpw .= $w;

  495.                                 if( strlen($tmpw) > 60 )

  496.                                 {

  497.                                     break;

  498.                                 }

  499.                             }

  500.                         }//while

  501.                         if( !$isok )

  502.                         {

  503.                             $this->simpleResult[$s]['w'] = $c;

  504.                               $this->simpleResult[$s]['t'] = 5;

  505.                               $s++;

  506.                               $onstr = '';

  507.                             $lastc = 5;

  508.                         }

  509.                         continue;

  510.                     }

  511.                     

  512.                     $onstr = '';

  513.                     $lastc = 5;

  514.                     if( $cn==0x3000 )

  515.                     {

  516.                         continue;

  517.                     }

  518.                     else

  519.                     {

  520.                         $this->simpleResult[$s]['w'] = $c;

  521.                         $this->simpleResult[$s]['t'] = 5;

  522.                         $s++;

  523.                     }

  524.                 }//2byte symbol

  525.                 

  526.             }//end 2byte char

  527.         

  528.         }//end for

  529.         

  530.         //处理分词后的结果

  531.         $this->_sort_finally_result();

  532.     }

  533.     

  534.     /**

  535.      * 深入分词

  536.      * @parem $str

  537.      * @parem $ctype (2 英文类， 3 中/韩/日文类)

  538.      * @parem $spos   当前粗分结果游标

  539.      * @return bool

  540.      */

  541.     function _deep_analysis( &$str, $ctype, $spos, $optimize=TRUE )

  542.     {

  543. 

  544.         //中文句子

  545.         if( $ctype==1 )

  546.         {

  547.             $slen = strlen($str);

  548.             //小于系统配置分词要求长度的句子

  549.             if( $slen < $this->notSplitLen )

  550.             {

  551.                 $tmpstr = '';

  552.                 $lastType = 0;

  553.                 if( $spos > 0 ) $lastType = $this->simpleResult[$spos-1]['t'];

  554.                 if($slen < 5)

  555.                 {

  556.                       //echo iconv(UCS2, 'utf-8', $str).'<br/>';

  557.                       if( $lastType==4 && ( isset($this->addonDic['u'][$str]) || isset($this->addonDic['u'][substr($str, 0, 2)]) ) )

  558.                       {

  559.                               $str2 = '';

  560.                               if( !isset($this->addonDic['u'][$str]) && isset($this->addonDic['s'][substr($str, 2, 2)]) )

  561.                               {

  562.                                      $str2 = substr($str, 2, 2);

  563.                                      $str  = substr($str, 0, 2);

  564.                               }

  565.                               $ww = $this->simpleResult[$spos - 1]['w'].$str;

  566.                               $this->simpleResult[$spos - 1]['w'] = $ww;

  567.                               $this->simpleResult[$spos - 1]['t'] = 4;

  568.                               if( !isset($this->newWords[$this->simpleResult[$spos - 1]['w']]) )

  569.                               {

  570.                                      $this->foundWordStr .= $this->_out_string_encoding( $ww ).'/mu, ';

  571.                                      $this->SetWordInfos($ww, array('c'=>1, 'm'=>'mu'));

  572.                               }

  573.                               $this->simpleResult[$spos]['w'] = '';

  574.                               if( $str2 != '' )

  575.                               {

  576.                                      $this->finallyResult[$spos-1][] = $ww;

  577.                                      $this->finallyResult[$spos-1][] = $str2;

  578.                               }

  579.                        }

  580.                        else {

  581.                               $this->finallyResult[$spos][] = $str;

  582.                        }

  583.                 }

  584.                 else

  585.                 {

  586.                       $this->_deep_analysis_cn( $str, $ctype, $spos, $slen, $optimize );

  587.                 }

  588.             }

  589.             //正常长度的句子，循环进行分词处理

  590.             else

  591.             {

  592.                 $this->_deep_analysis_cn( $str, $ctype, $spos, $slen, $optimize );

  593.             }

  594.         }

  595.         //英文句子，转为小写

  596.         else

  597.         {

  598.             if( $this->toLower ) {

  599.                 $this->finallyResult[$spos][] = strtolower($str);

  600.             }

  601.             else {

  602.                 $this->finallyResult[$spos][] = $str;

  603.             }

  604.         }

  605.     }

  606.     

  607.     /**

  608.      * 中文的深入分词

  609.      * @parem $str

  610.      * @return void

  611.      */

  612.     function _deep_analysis_cn( &$str, $lastec, $spos, $slen, $optimize=TRUE )

  613.     {

  614.         $quote1 = chr(0x20).chr(0x1C);

  615.         $tmparr = array();

  616.         $hasw = 0;

  617.         //如果前一个词为 “ ， 并且字符串小于3个字符当成一个词处理。

  618.         if( $spos > 0 && $slen < 11 && $this->simpleResult[$spos-1]['w']==$quote1 )

  619.         {

  620.             $tmparr[] = $str;

  621.             if( !isset($this->newWords[$str]) )

  622.             {

  623.                 $this->foundWordStr .= $this->_out_string_encoding($str).'/nq, ';

  624.                 $this->SetWordInfos($str, array('c'=>1, 'm'=>'nq'));

  625.             }

  626.             if( !$this->differMax )

  627.             {

  628.                 $this->finallyResult[$spos][] = $str;

  629.                 return ;

  630.             }

  631.         }

  632.         //进行切分

  633.         for($i=$slen-1; $i > 0; $i -= 2)

  634.         {

  635.             //单个词

  636.             $nc = $str[$i-1].$str[$i];

  637.             //是否已经到最后两个字

  638.             if( $i <= 2 )

  639.             {

  640.                 $tmparr[] = $nc;

  641.                 $i = 0;

  642.                 break;

  643.             }

  644.             $isok = FALSE;

  645.             $i = $i + 1;

  646.             for($k=$this->dicWordMax; $k>1; $k=$k-2)

  647.             {

  648.                 if($i < $k) continue;

  649.                 $w = substr($str, $i-$k, $k);

  650.                 if( strlen($w) <= 2 )

  651.                 {

  652.                     $i = $i - 1;

  653.                     break;

  654.                 }

  655.                 if( $this->IsWord( $w ) )

  656.                 {

  657.                     $tmparr[] = $w;

  658.                     $i = $i - $k + 1;

  659.                     $isok = TRUE;

  660.                     break;

  661.                 }

  662.             }

  663.             //echo '<hr />';

  664.             //没适合词

  665.             if(!$isok) $tmparr[] = $nc;

  666.         }

  667.         $wcount = count($tmparr);

  668.         if( $wcount==0 ) return ;

  669.         $this->finallyResult[$spos] = array_reverse($tmparr);

  670.         //优化结果(岐义处理、新词、数词、人名识别等)

  671.         if( $optimize )

  672.         {

  673.             $this->_optimize_result( $this->finallyResult[$spos], $spos );

  674.         }

  675.     }

  676.     

  677.     /**

  678.     * 对最终分词结果进行优化（把simpleresult结果合并，并尝试新词识别、数词合并等）

  679.     * @parem $optimize 是否优化合并的结果

  680.     * @return bool

  681.     */

  682.     //t = 1 中/韩/日文, 2 英文/数字/符号('.', '@', '#', '+'), 3 ANSI符号 4 纯数字 5 非ANSI符号或不支持字符

  683.     function _optimize_result( &$smarr, $spos )

  684.     {

  685.         $newarr = array();

  686.         $prePos = $spos - 1;

  687.         $arlen = count($smarr);

  688.         $i = $j = 0;

  689.         //检测数量词

  690.         if( $prePos > -1 && !isset($this->finallyResult[$prePos]) )

  691.         {

  692.             $lastw = $this->simpleResult[$prePos]['w'];

  693.             $lastt = $this->simpleResult[$prePos]['t'];

  694.               if( ($lastt==4 || isset( $this->addonDic['c'][$lastw] )) && isset( $this->addonDic['u'][$smarr[0]] ) )

  695.               {

  696.                  $this->simpleResult[$prePos]['w'] = $lastw.$smarr[0];

  697.                  $this->simpleResult[$prePos]['t'] = 4;

  698.                  if( !isset($this->newWords[ $this->simpleResult[$prePos]['w'] ]) )

  699.                  {

  700.                      $this->foundWordStr .= $this->_out_string_encoding( $this->simpleResult[$prePos]['w'] ).'/mu, ';

  701.                      $this->SetWordInfos($this->simpleResult[$prePos]['w'], array('c'=>1, 'm'=>'mu'));

  702.                  }

  703.                  $smarr[0] = '';

  704.                  $i++;

  705.               }

  706.        }

  707.        for(; $i < $arlen; $i++)

  708.        {

  709.             

  710.             if( !isset( $smarr[$i+1] ) )

  711.             {

  712.                 $newarr[$j] = $smarr[$i];

  713.                 break;

  714.             }

  715.             $cw = $smarr[$i];

  716.             $nw = $smarr[$i+1];

  717.             $ischeck = FALSE;

  718.             //检测数量词

  719.             if( isset( $this->addonDic['c'][$cw] ) && isset( $this->addonDic['u'][$nw] ) )

  720.             {

  721.                 //最大切分时保留合并前的词

  722.                 if($this->differMax)

  723.                 {

  724.                         $newarr[$j] = chr(0).chr(0x28);

  725.                         $j++;

  726.                         $newarr[$j] = $cw;

  727.                         $j++;

  728.                         $newarr[$j] = $nw;

  729.                         $j++;

  730.                         $newarr[$j] = chr(0).chr(0x29);

  731.                         $j++;

  732.                 }

  733.                 $newarr[$j] = $cw.$nw;

  734.                 if( !isset($this->newWords[$newarr[$j]]) )

  735.                 {

  736.                     $this->foundWordStr .= $this->_out_string_encoding( $newarr[$j] ).'/mu, ';

  737.                     $this->SetWordInfos($newarr[$j], array('c'=>1, 'm'=>'mu'));

  738.                 }

  739.                 $j++; $i++; $ischeck = TRUE;

  740.             }

  741.             //检测前导词(通常是姓)

  742.             else if( isset( $this->addonDic['n'][ $smarr[$i] ] ) )

  743.             {

  744.                 $is_rs = FALSE;

  745.                 //词语是副词或介词或频率很高的词不作为人名

  746.                 if( strlen($nw)==4 )

  747.                 {

  748.                     $winfos = $this->GetWordInfos($nw);

  749.                     if(isset($winfos['m']) && ($winfos['m']=='r' || $winfos['m']=='c' || $winfos['c']>500) )

  750.                     {

  751.                          $is_rs = TRUE;

  752.                     }

  753.                 }

  754.                 if( !isset($this->addonDic['s'][$nw]) && strlen($nw)<5 && !$is_rs )

  755.                 {

  756.                     $newarr[$j] = $cw.$nw;

  757.                     //echo iconv(UCS2, 'utf-8', $newarr[$j])."<br />";

  758.                     //尝试检测第三个词

  759.                     if( strlen($nw)==2 && isset($smarr[$i+2]) && strlen($smarr[$i+2])==2 && !isset( $this->addonDic['s'][$smarr[$i+2]] ) )

  760.                     {

  761.                         $newarr[$j] .= $smarr[$i+2];

  762.                         $i++;

  763.                     }

  764.                     if( !isset($this->newWords[$newarr[$j]]) )

  765.                     {

  766.                         $this->SetWordInfos($newarr[$j], array('c'=>1, 'm'=>'nr'));

  767.                         $this->foundWordStr .= $this->_out_string_encoding($newarr[$j]).'/nr, ';

  768.                     }

  769.                     //为了防止错误，保留合并前的姓名

  770.                     if(strlen($nw)==4)

  771.                     {

  772.                         $j++;

  773.                         $newarr[$j] = chr(0).chr(0x28);

  774.                         $j++;

  775.                         $newarr[$j] = $cw;

  776.                         $j++;

  777.                         $newarr[$j] = $nw;

  778.                         $j++;

  779.                         $newarr[$j] = chr(0).chr(0x29);

  780.                     }

  781.                     

  782.                     $j++; $i++; $ischeck = TRUE;

  783.                 }

  784.             }

  785.             //检测后缀词(地名等)

  786.             else if( isset($this->addonDic['a'][$nw]) )

  787.             {

  788.                 $is_rs = FALSE;

  789.                 //词语是副词或介词不作为前缀

  790.                 if( strlen($cw)>2 )

  791.                 {

  792.                     $winfos = $this->GetWordInfos($cw);

  793.                     if(isset($winfos['m']) && ($winfos['m']=='a' || $winfos['m']=='r' || $winfos['m']=='c' || $winfos['c']>500) )

  794.                     {

  795.                          $is_rs = TRUE;

  796.                     }

  797.                 }

  798.                 if( !isset($this->addonDic['s'][$cw]) && !$is_rs )

  799.                 {

  800.                     $newarr[$j] = $cw.$nw;

  801.                     if( !isset($this->newWords[$newarr[$j]]) )

  802.                     {

  803.                         $this->foundWordStr .= $this->_out_string_encoding($newarr[$j]).'/na, ';

  804.                         $this->SetWordInfos($newarr[$j], array('c'=>1, 'm'=>'na'));

  805.                     }

  806.                     $i++; $j++; $ischeck = TRUE;

  807.                 }

  808.             }

  809.             //新词识别（暂无规则）

  810.             else if($this->unitWord)

  811.             {

  812.                 if(strlen($cw)==2 && strlen($nw)==2 

  813.                 && !isset($this->addonDic['s'][$cw]) && !isset($this->addonDic['t'][$cw]) && !isset($this->addonDic['a'][$cw]) 

  814.                 && !isset($this->addonDic['s'][$nw]) && !isset($this->addonDic['c'][$nw]))

  815.                 {

  816.                     $newarr[$j] = $cw.$nw;

  817.                     //尝试检测第三个词

  818.                     if( isset($smarr[$i+2]) && strlen($smarr[$i+2])==2 && (isset( $this->addonDic['a'][$smarr[$i+2]] ) || isset( $this->addonDic['u'][$smarr[$i+2]] )) )

  819.                     {

  820.                         $newarr[$j] .= $smarr[$i+2];

  821.                         $i++;

  822.                     }

  823.                     if( !isset($this->newWords[$newarr[$j]]) )

  824.                     {

  825.                         $this->foundWordStr .= $this->_out_string_encoding($newarr[$j]).'/ms, ';

  826.                         $this->SetWordInfos($newarr[$j], array('c'=>1, 'm'=>'ms'));

  827.                     }

  828.                     $i++; $j++; $ischeck = TRUE;

  829.                 }

  830.             }

  831.             

  832.             //不符合规则

  833.             if( !$ischeck )

  834.             {

  835.                 $newarr[$j] = $cw;

  836.                   //二元消岐处理——最大切分模式

  837.                 if( $this->differMax && !isset($this->addonDic['s'][$cw]) && strlen($cw) < 5 && strlen($nw) < 7)

  838.                 {

  839.                     $slen = strlen($nw);

  840.                     $hasDiff = FALSE;

  841.                     for($y=2; $y <= $slen-2; $y=$y+2)

  842.                     {

  843.                         $nhead = substr($nw, $y-2, 2);

  844.                         $nfont = $cw.substr($nw, 0, $y-2);

  845.                         if( $this->IsWord( $nfont.$nhead ) )

  846.                         {

  847.                             if( strlen($cw) > 2 ) $j++;

  848.                             $hasDiff = TRUE;

  849.                             $newarr[$j] = $nfont.$nhead;

  850.                         }

  851.                     }

  852.                 }

  853.                 $j++;

  854.             }

  855.             

  856.        }//end for

  857.        $smarr =  $newarr;

  858.     }

  859.     

  860.     /**

  861.     * 转换最终分词结果到 finallyResult 数组

  862.     * @return void

  863.     */

  864.     function _sort_finally_result()

  865.     {

  866.           $newarr = array();

  867.         $i = 0;

  868.         foreach($this->simpleResult as $k=>$v)

  869.         {

  870.             if( empty($v['w']) ) continue;

  871.             if( isset($this->finallyResult[$k]) && count($this->finallyResult[$k]) > 0 )

  872.             {

  873.                 foreach($this->finallyResult[$k] as $w)

  874.                 {

  875.                     if(!empty($w))

  876.                     {

  877.                         $newarr[$i]['w'] = $w;

  878.                         $newarr[$i]['t'] = 20;

  879.                         $i++;

  880.                     }

  881.                 }

  882.             }

  883.             else if($v['t'] != 21)

  884.             {

  885.                 $newarr[$i]['w'] = $v['w'];

  886.                 $newarr[$i]['t'] = $v['t'];

  887.                 $i++;

  888.             }

  889.         }

  890.         $this->finallyResult = $newarr;

  891.         $newarr = '';

  892.       }

  893.     

  894.     /**

  895.      * 把uncode字符串转换为输出字符串

  896.      * @parem str

  897.      * return string

  898.      */

  899.      function _out_string_encoding( &$str )

  900.      {

  901.         $rsc = $this->_source_result_charset();

  902.         if( $rsc==1 ) {

  903.             $rsstr = iconv(UCS2, 'utf-8', $str);

  904.         }

  905.         else if( $rsc==2 ) {

  906.             $rsstr = iconv('utf-8', 'gb18030', iconv(UCS2, 'utf-8', $str) );

  907.         }

  908.         else{

  909.             $rsstr = iconv('utf-8', 'big5', iconv(UCS2, 'utf-8', $str) );

  910.         }

  911.         return $rsstr;

  912.      }

  913.     

  914.     /**

  915.      * 获取最终结果字符串（用空格分开后的分词结果）

  916.      * @return string

  917.      */

  918.      function GetFinallyResult($spword=' ', $word_meanings=FALSE)

  919.      {

  920.         $rsstr = '';

  921.         foreach($this->finallyResult as $v)

  922.         {

  923.             if( $this->resultType==2 && ($v['t']==3 || $v['t']==5) )

  924.             {

  925.                 continue;

  926.             }

  927.             $m = '';

  928.             if( $word_meanings )

  929.             {

  930.                 $m = $this->GetWordProperty($v['w']);

  931.             }

  932.             $w = $this->_out_string_encoding($v['w']);

  933.             if( $w != ' ' )

  934.             {

  935.                 if($word_meanings) {

  936.                     $rsstr .= $spword.$w.$m;

  937.                 }

  938.                 else {

  939.                     $rsstr .= $spword.$w;

  940.                 }

  941.             }

  942.         }

  943.         return $rsstr;

  944.      }

  945.      

  946.     /**

  947.      * 获取粗分结果，不包含粗分属性

  948.      * @return array()

  949.      */

  950.      function GetSimpleResult()

  951.      {

  952.         $rearr = array();

  953.         foreach($this->simpleResult as $k=>$v)

  954.         {

  955.             if( empty($v['w']) ) continue;

  956.             $w = $this->_out_string_encoding($v['w']);

  957.             if( $w != ' ' ) $rearr[] = $w;

  958.         }

  959.         return $rearr;

  960.      }

  961.      

  962.     /**

  963.      * 获取粗分结果，包含粗分属性（1中文词句、2 ANSI词汇（包括全角），3 ANSI标点符号（包括全角），4数字（包括全角），5 中文标点或无法识别字符）

  964.      * @return array()

  965.      */

  966.      function GetSimpleResultAll()

  967.      {

  968.         $rearr = array();

  969.         foreach($this->simpleResult as $k=>$v)

  970.         {

  971.             $w = $this->_out_string_encoding($v['w']);

  972.             if( $w != ' ' )

  973.             {

  974.                 $rearr[$k]['w'] = $w;

  975.                 $rearr[$k]['t'] = $v['t'];

  976.             }

  977.         }

  978.         return $rearr;

  979.      }

  980.      

  981.     /**

  982.      * 获取索引hash数组

  983.      * @return array('word'=>count,...)

  984.      */

  985.      function GetFinallyIndex()

  986.      {

  987.         $rearr = array();

  988.         foreach($this->finallyResult as $v)

  989.         {

  990.             if( $this->resultType==2 && ($v['t']==3 || $v['t']==5) )

  991.             {

  992.                 continue;

  993.             }

  994.             $w = $this->_out_string_encoding($v['w']);

  995.             if( $w == ' ' )

  996.             {

  997.                 continue;

  998.             }

  999.             if( isset($rearr[$w]) )

  1000.             {

  1001.                  $rearr[$w]++;

  1002.             }

  1003.             else

  1004.             {

  1005.                  $rearr[$w] = 1;

  1006.             }

  1007.         }

  1008.         return $rearr;

  1009.      }

  1010.      

  1011.     /**

  1012.      * 获得保存目标编码

  1013.      * @return int

  1014.      */

  1015.      function _source_result_charset()

  1016.      {

  1017.         if( preg_match("/^utf/", $this->targetCharSet) ) {

  1018.            $rs = 1;

  1019.         }

  1020.         else if( preg_match("/^gb/", $this->targetCharSet) ) {

  1021.            $rs = 2;

  1022.         }

  1023.         else if( preg_match("/^big/", $this->targetCharSet) ) {

  1024.            $rs = 3;

  1025.         }

  1026.         else {

  1027.             $rs = 4;

  1028.         }

  1029.         return $rs;

  1030.      }

  1031.      

  1032.      /**

  1033.      * 编译词典

  1034.      * @parem $sourcefile utf-8编码的文本词典数据文件<参见范例dict/not-build/base_dic_full.txt>

  1035.      * 注意, 需要PHP开放足够的内存才能完成操作

  1036.      * @return void

  1037.      */

  1038.      function MakeDict( $source_file, $target_file='' )

  1039.      {

  1040.         $target_file = ($target_file=='' ? $this->mainDicFile : $target_file);

  1041.         $allk = array();

  1042.         $fp = fopen($source_file, 'r');

  1043.         while( $line = fgets($fp, 512) )

  1044.         {

  1045.             if( $line[0]=='@' ) continue;

  1046.             list($w, $r, $a) = explode(',', $line);

  1047.             $a = trim( $a );

  1048.             $w = iconv('utf-8', UCS2, $w);

  1049.             $k = $this->_get_index( $w );

  1050.             if( isset($allk[ $k ]) )

  1051.                 $allk[ $k ][ $w ] = array($r, $a);

  1052.             else

  1053.                 $allk[ $k ][ $w ] = array($r, $a);

  1054.         }

  1055.         fclose( $fp );

  1056.         $fp = fopen($target_file, 'w');

  1057.         $heade_rarr = array();

  1058.         $alldat = '';

  1059.         $start_pos = $this->mask_value * 8;

  1060.         foreach( $allk as $k => $v )

  1061.         {

  1062.             $dat  = serialize( $v );

  1063.             $dlen = strlen($dat);

  1064.             $alldat .= $dat;

  1065.         

  1066.             $heade_rarr[ $k ][0] = $start_pos;

  1067.             $heade_rarr[ $k ][1] = $dlen;

  1068.             $heade_rarr[ $k ][2] = count( $v );

  1069.         

  1070.             $start_pos += $dlen;

  1071.         }

  1072.         unset( $allk );

  1073.         for($i=0; $i < $this->mask_value; $i++)

  1074.         {

  1075.             if( !isset($heade_rarr[$i]) )

  1076.             {

  1077.                 $heade_rarr[$i] = array(0, 0, 0);

  1078.             }

  1079.             fwrite($fp, pack("Inn", $heade_rarr[$i][0], $heade_rarr[$i][1], $heade_rarr[$i][2]));

  1080.         }

  1081.         fwrite( $fp, $alldat);

  1082.         fclose( $fp );

  1083.      }

  1084.      

  1085.      /**

  1086.      * 导出词典的词条

  1087.      * @parem $targetfile 保存位置

  1088.      * @return void

  1089.      */

  1090.      function ExportDict( $targetfile )

  1091.      {

  1092.         if( !$this->mainDicHand )

  1093.         {

  1094.             $this->mainDicHand = fopen($this->mainDicFile, 'rw');

  1095.         }

  1096.         $fp = fopen($targetfile, 'w');

  1097.         for($i=0; $i <= $this->mask_value; $i++)

  1098.         {

  1099.             $move_pos = $i * 8;

  1100.             fseek($this->mainDicHand, $move_pos, SEEK_SET);

  1101.             $dat = fread($this->mainDicHand, 8);

  1102.             $arr = unpack('I1s/n1l/n1c', $dat);

  1103.             if( $arr['l'] == 0 )

  1104.             {

  1105.                 continue;

  1106.             }

  1107.             fseek($this->mainDicHand, $arr['s'], SEEK_SET);

  1108.             $data = @unserialize(fread($this->mainDicHand, $arr['l']));

  1109.             if( !is_array($data) ) continue;

  1110.             foreach($data as $k => $v)

  1111.             {

  1112.                 $w = iconv(UCS2, 'utf-8', $k);

  1113.                 fwrite($fp, "{$w},{$v[0]},{$v[1]}\n");

  1114.             }

  1115.         }

  1116.         fclose( $fp );

  1117.         return TRUE;

  1118.      }

  1119.      

  1120. 	function InportDict( $targetfile )

  1121.      {

  1122.      	if(!ini_set('memory_limit', '128M'))

  1123. 			exit('设置内存错误，请到dede官网下载解压版的base_dic_full.dic!');

  1124.      	require_once(DEDEINC.'/zip.class.php');

  1125.      	$zip = new zip();

  1126.      	//echo $targetfile;

  1127.      	$unpackagefile = array_keys($zip->Extract($targetfile,DEDEINC.'/data/'));

  1128.      	//exit();

  1129.      	$this->MakeDict(DEDEINC.'/data/'.$unpackagefile[0]);

  1130.      	unlink(DEDEINC.'/data/'.$unpackagefile[0]);

  1131.      	return true;

  1132.      }

  1133. }