DedeBIZ
/
DedeV6
Watch 2
Star 0
Fork 0
Code Pull Requests 0 Releases 29 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.
449 Commits
3 Branches
322MB
Tree: 776f65b64b
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '776f65b64b'
${ noResults }
DedeV6/src/include/splitword.class.php

1125 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) 2022, DedeBIZ.COM

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

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

  15.  */

  16. //常量定义

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

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

  19. class SplitWord

  20. {

  21.     

  22.     //hash算法选项

  23.     var $mask_value = 0xFFFF;

  24.     

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

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

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

  28.     

  29.     //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

  30.     var $resultType = 1;

  31.     

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

  33.     var $notSplitLen = 5;

  34.     

  35.     //把英文单词全部转小写

  36.     var $toLower = FALSE;

  37.     

  38.     //使用最大切分模式对二元词进行消岐

  39.     var $differMax = FALSE;

  40.     

  41.     //尝试合并单字

  42.     var $unitWord = TRUE;

  43.     

  44.     //初始化类时直接加载词典

  45.     var $loadInit = TRUE;

  46.     

  47.     //使用热门词优先模式进行消岐

  48.     var $differFreq = FALSE;

  49.     

  50.     //被转换为unicode的源字符串

  51.     var $sourceString = '';

  52.     

  53.     //附加词典

  54.     var $addonDic = array();

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

  56.     

  57.     //主词典 

  58.     var $dicStr = '';

  59.     var $mainDic = array();

  60.     var $mainDicHand = FALSE;

  61.     var $mainDicInfos = array();

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

  63.     //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）

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

  65.     var $isLoadAll = FALSE;

  66.     var $isUnpacked = FALSE;

  67.     

  68.     //主词典词语最大长度 x / 2

  69.     var $dicWordMax = 14;

  70.     //粗分后的数组（通常是截取句子等用途）

  71.     var $simpleResult = array();

  72.     //最终结果(用空格分开的词汇列表)

  73.     var $finallyResult = '';

  74.     

  75.     //是否已经载入词典

  76.     var $isLoadDic = FALSE;

  77.     //系统识别或合并的新词

  78.     var $newWords = array();

  79.     var $foundWordStr = '';

  80.     //词库载入时间

  81.     var $loadTime = 0;

  82.     

  83.     /**

  84.      * 构造函数

  85.      * @param $source_charset

  86.      * @param $target_charset

  87.      * @param $load_alldic 

  88.      * @param $source

  89.      *

  90.      * @return void

  91.      */

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

  93.     {

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

  95.         $this->isLoadAll = $load_all;

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

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

  98.     }

  99.     

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

  101.     {

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

  103.     }

  104.     

  105.    /**

  106.     * 析构函数

  107.     */

  108.     function __destruct()

  109.     {

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

  111.         {

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

  113.         }

  114.     }

  115.     

  116.     /**

  117.      * 根据字符串计算key索引

  118.      * @param $key

  119.      * @return short int

  120.      */

  121.     function _get_index( $key )

  122.     {

  123.         $l = strlen($key);

  124.         $h = 0x238f13af;

  125.         while ($l--)

  126.         {

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

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

  129.             $h &= 0x7fffffff;

  130.         }

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

  132.     }

  133.     

  134.     /**

  135.      * 从文件获得词

  136.      * @param $key

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

  138.      * @return short int

  139.      */

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

  141.     {

  142.         if( !$this->mainDicHand )

  143.         {

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

  145.         }

  146.         $p = 0;

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

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

  149.         {

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

  151.         } else {

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

  153.             $move_pos = $keynum * 8;

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

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

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

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

  158.             {

  159.                 return FALSE;

  160.             }

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

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

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

  164.        }

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

  166.        {

  167.            return FALSE;

  168.        }

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

  170.     }

  171.     

  172.     /**

  173.      * 设置源字符串

  174.      * @param $source

  175.      * @param $source_charset

  176.      * @param $target_charset

  177.      *

  178.      * @return bool

  179.      */

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

  181.     {

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

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

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

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

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

  187.         if( $source != '' )

  188.         {

  189.             $rs = TRUE;

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

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

  192.             }

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

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

  195.             }

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

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

  198.             }

  199.             else {

  200.                 $rs = FALSE;

  201.             }

  202.         } else {

  203.            $rs = FALSE;

  204.         }

  205.         return $rs;

  206.     }

  207.     

  208.     /**

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

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

  211.      *

  212.      * @return void

  213.      */

  214.     function SetResultType( $rstype )

  215.     {

  216.         $this->resultType = $rstype;

  217.     }

  218.     

  219.     /**

  220.      * 载入词典

  221.      *

  222.      * @return void

  223.      */

  224.     function LoadDict( $maindic='' )

  225.     {

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

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

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

  229.         $startt = microtime(TRUE);

  230.         //正常读取文件

  231.         $dicAddon = $this->addonDicFile;

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

  233.         {

  234.             $dicWords = $this->mainDicFile ;

  235.         } else {

  236.             $dicWords = $maindic;

  237.             $this->mainDicFile = $maindic;

  238.         }

  239.         

  240.         //加载主词典（只打开）

  241.         if($this->isUnpacked){

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

  243.         } else {

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

  245.         }

  246.         

  247.         //载入副词典

  248.         $hw = '';

  249.         $ds = file($dicAddon);

  250.         foreach($ds as $d)

  251.         {

  252.             $d = trim($d);

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

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

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

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

  257.             }

  258.             else

  259.             {

  260.                 $spstr = _SP_;

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

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

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

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

  265.                 foreach($ws as $estr)

  266.                 {

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

  268.                 }

  269.             }

  270.         }

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

  272.         $this->isLoadDic = TRUE;

  273.     }

  274.     

  275.    /**

  276.     * 检测某个词是否存在

  277.     */

  278.     function IsWord( $word )

  279.     {

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

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

  282.     }

  283.     

  284.     /**

  285.      * 获得某个词的词性及词频信息

  286.      * @parem $word unicode编码的词

  287.      * @return void

  288.      */

  289.      function GetWordProperty($word)

  290.      {

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

  292.         {

  293.             return '/s';

  294.         }

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

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

  297.      }

  298.     

  299.     /**

  300.      * 指定某词的词性信息（通常是新词）

  301.      * @parem $word unicode编码的词

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

  303.      * @return void;

  304.      */

  305.     function SetWordInfos($word, $infos)

  306.     {

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

  308.         {

  309.             return ;

  310.         }

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

  312.         {

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

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

  315.         } else {

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

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

  318.         }

  319.     }

  320.     

  321.     /**

  322.      * 开始执行分析

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

  324.      * @return bool

  325.      */

  326.     function StartAnalysis($optimize=TRUE)

  327.     {

  328.         if( !$this->isLoadDic )

  329.         {

  330.             $this->LoadDict();

  331.         }

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

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

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

  335.         $sbcArr = array();

  336.         $j = 0;

  337.         //全角与半角字符对照表

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

  339.         {

  340.             $scb = 0x20 + $j;

  341.             $j++;

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

  343.         }

  344.         //对字符串进行粗分

  345.         $onstr = '';

  346.         $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

  347.         $s = 0;

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

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

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

  351.         {

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

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

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

  355.             //ANSI字符

  356.             if($cn < 0x80)

  357.             {

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

  359.                 {

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

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

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

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

  364.                         $s++;

  365.                         $onstr = '';

  366.                     }

  367.                     $lastc = 2;

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

  369.                 }

  370.                 else

  371.                 {

  372.                     if( $onstr != '' )

  373.                     {

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

  375.                         if( $lastc==2 )

  376.                         {

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

  378.                         }

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

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

  381.                         $s++;

  382.                     }

  383.                     $onstr = '';

  384.                     $lastc = 3;

  385.                     if($cn < 31)

  386.                     {

  387.                         continue;

  388.                     }

  389.                     else

  390.                     {

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

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

  393.                         $s++;

  394.                     }

  395.                 }

  396.             }

  397.             //普通字符

  398.             else

  399.             {

  400.                 //正常文字

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

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

  403.                 {

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

  405.                     {

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

  407.                         if( $lastc==2 )

  408.                         {

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

  410.                         }

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

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

  413.                         $s++;

  414.                         $onstr = '';

  415.                     }

  416.                     $lastc = 1;

  417.                     $onstr .= $c;

  418.                 }

  419.                 //特殊符号

  420.                 else

  421.                 {

  422.                     if( $onstr != '' )

  423.                     {

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

  425.                         if( $lastc==2 )

  426.                         {

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

  428.                         }

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

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

  431.                         $s++;

  432.                     }

  433.                     

  434.                     //检测书名

  435.                     if( $cn == 0x300A )

  436.                     {

  437.                         $tmpw = '';

  438.                         $n = 1;

  439.                         $isok = FALSE;

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

  441.                         while(TRUE)

  442.                         {

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

  444. 							break;

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

  446.                             if( $w == $ew )

  447.                             {

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

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

  450.                                 $s++;

  451.                         

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

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

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

  455.                                 {

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

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

  458.                                 }

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

  460.                                 

  461.                                 $s++;

  462. 

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

  464.                                 if( $this->differMax )

  465.                                 {

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

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

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

  469.                                     $s++;

  470.                                 }

  471.                                 

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

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

  474.                                 $s++;

  475.                         

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

  477.                                 $isok = TRUE;

  478.                                 $onstr = '';

  479.                                 $lastc = 5;

  480.                                 break;

  481.                             }

  482.                             else

  483.                             {

  484.                                 $n = $n+2;

  485.                                 $tmpw .= $w;

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

  487.                                 {

  488.                                     break;

  489.                                 }

  490.                             }

  491.                         }//while

  492.                         if( !$isok )

  493.                         {

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

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

  496.                               $s++;

  497.                               $onstr = '';

  498.                             $lastc = 5;

  499.                         }

  500.                         continue;

  501.                     }

  502.                     

  503.                     $onstr = '';

  504.                     $lastc = 5;

  505.                     if( $cn==0x3000 )

  506.                     {

  507.                         continue;

  508.                     }

  509.                     else

  510.                     {

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

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

  513.                         $s++;

  514.                     }

  515.                 }//2byte symbol

  516.                 

  517.             }//end 2byte char

  518.         

  519.         }//end for

  520.         

  521.         //处理分词后的结果

  522.         $this->_sort_finally_result();

  523.     }

  524.     

  525.     /**

  526.      * 深入分词

  527.      * @parem $str

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

  529.      * @parem $spos   当前粗分结果游标

  530.      * @return bool

  531.      */

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

  533.     {

  534. 

  535.         //中文句子

  536.         if( $ctype==1 )

  537.         {

  538.             $slen = strlen($str);

  539.             //小于系统配置分词要求长度的句子

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

  541.             {

  542.                 $tmpstr = '';

  543.                 $lastType = 0;

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

  545.                 if($slen < 5)

  546.                 {

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

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

  549.                       {

  550.                               $str2 = '';

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

  552.                               {

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

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

  555.                               }

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

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

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

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

  560.                               {

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

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

  563.                               }

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

  565.                               if( $str2 != '' )

  566.                               {

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

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

  569.                               }

  570.                        }

  571.                        else {

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

  573.                        }

  574.                 }

  575.                 else

  576.                 {

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

  578.                 }

  579.             }

  580.             //正常长度的句子，循环进行分词处理

  581.             else

  582.             {

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

  584.             }

  585.         }

  586.         //英文句子，转为小写

  587.         else

  588.         {

  589.             if( $this->toLower ) {

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

  591.             }

  592.             else {

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

  594.             }

  595.         }

  596.     }

  597.     

  598.     /**

  599.      * 中文的深入分词

  600.      * @parem $str

  601.      * @return void

  602.      */

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

  604.     {

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

  606.         $tmparr = array();

  607.         $hasw = 0;

  608.         //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。

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

  610.         {

  611.             $tmparr[] = $str;

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

  613.             {

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

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

  616.             }

  617.             if( !$this->differMax )

  618.             {

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

  620.                 return ;

  621.             }

  622.         }

  623.         //进行切分

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

  625.         {

  626.             //单个词

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

  628.             //是否已经到最后两个字

  629.             if( $i <= 2 )

  630.             {

  631.                 $tmparr[] = $nc;

  632.                 $i = 0;

  633.                 break;

  634.             }

  635.             $isok = FALSE;

  636.             $i = $i + 1;

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

  638.             {

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

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

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

  642.                 {

  643.                     $i = $i - 1;

  644.                     break;

  645.                 }

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

  647.                 {

  648.                     $tmparr[] = $w;

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

  650.                     $isok = TRUE;

  651.                     break;

  652.                 }

  653.             }

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

  655.             //没适合词

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

  657.         }

  658.         $wcount = count($tmparr);

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

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

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

  662.         if( $optimize )

  663.         {

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

  665.         }

  666.     }

  667.     

  668.     /**

  669.     * 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）

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

  671.     * @return bool

  672.     */

  673.     //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

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

  675.     {

  676.         $newarr = array();

  677.         $prePos = $spos - 1;

  678.         $arlen = count($smarr);

  679.         $i = $j = 0;

  680.         //检测数量词

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

  682.         {

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

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

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

  686.               {

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

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

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

  690.                  {

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

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

  693.                  }

  694.                  $smarr[0] = '';

  695.                  $i++;

  696.               }

  697.        }

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

  699.        {

  700.             

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

  702.             {

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

  704.                 break;

  705.             }

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

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

  708.             $ischeck = FALSE;

  709.             //检测数量词

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

  711.             {

  712.                 //最大切分时保留合并前的词

  713.                 if($this->differMax)

  714.                 {

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

  716.                         $j++;

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

  718.                         $j++;

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

  720.                         $j++;

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

  722.                         $j++;

  723.                 }

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

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

  726.                 {

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

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

  729.                 }

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

  731.             }

  732.             //检测前导词(通常是姓)

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

  734.             {

  735.                 $is_rs = FALSE;

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

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

  738.                 {

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

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

  741.                     {

  742.                          $is_rs = TRUE;

  743.                     }

  744.                 }

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

  746.                 {

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

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

  749.                     //尝试检测第三个词

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

  751.                     {

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

  753.                         $i++;

  754.                     }

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

  756.                     {

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

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

  759.                     }

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

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

  762.                     {

  763.                         $j++;

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

  765.                         $j++;

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

  767.                         $j++;

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

  769.                         $j++;

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

  771.                     }

  772.                     

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

  774.                 }

  775.             }

  776.             //检测后缀词(地名等)

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

  778.             {

  779.                 $is_rs = FALSE;

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

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

  782.                 {

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

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

  785.                     {

  786.                          $is_rs = TRUE;

  787.                     }

  788.                 }

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

  790.                 {

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

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

  793.                     {

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

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

  796.                     }

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

  798.                 }

  799.             }

  800.             //新词识别（暂无规则）

  801.             else if($this->unitWord)

  802.             {

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

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

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

  806.                 {

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

  808.                     //尝试检测第三个词

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

  810.                     {

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

  812.                         $i++;

  813.                     }

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

  815.                     {

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

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

  818.                     }

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

  820.                 }

  821.             }

  822.             

  823.             //不符合规则

  824.             if( !$ischeck )

  825.             {

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

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

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

  829.                 {

  830.                     $slen = strlen($nw);

  831.                     $hasDiff = FALSE;

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

  833.                     {

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

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

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

  837.                         {

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

  839.                             $hasDiff = TRUE;

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

  841.                         }

  842.                     }

  843.                 }

  844.                 $j++;

  845.             }

  846.             

  847.        }//end for

  848.        $smarr =  $newarr;

  849.     }

  850.     

  851.     /**

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

  853.     * @return void

  854.     */

  855.     function _sort_finally_result()

  856.     {

  857.           $newarr = array();

  858.         $i = 0;

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

  860.         {

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

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

  863.             {

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

  865.                 {

  866.                     if(!empty($w))

  867.                     {

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

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

  870.                         $i++;

  871.                     }

  872.                 }

  873.             }

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

  875.             {

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

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

  878.                 $i++;

  879.             }

  880.         }

  881.         $this->finallyResult = $newarr;

  882.         $newarr = '';

  883.       }

  884.     

  885.     /**

  886.      * 把uncode字符串转换为输出字符串

  887.      * @parem str

  888.      * return string

  889.      */

  890.      function _out_string_encoding( &$str )

  891.      {

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

  893.         if( $rsc==1 ) {

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

  895.         }

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

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

  898.         }

  899.         else{

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

  901.         }

  902.         return $rsstr;

  903.      }

  904.     

  905.     /**

  906.      * 获取最终结果字符串（用空格分开后的分词结果）

  907.      * @return string

  908.      */

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

  910.      {

  911.         $rsstr = '';

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

  913.         {

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

  915.             {

  916.                 continue;

  917.             }

  918.             $m = '';

  919.             if( $word_meanings )

  920.             {

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

  922.             }

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

  924.             if( $w != ' ' )

  925.             {

  926.                 if($word_meanings) {

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

  928.                 }

  929.                 else {

  930.                     $rsstr .= $spword.$w;

  931.                 }

  932.             }

  933.         }

  934.         return $rsstr;

  935.      }

  936.      

  937.     /**

  938.      * 获取粗分结果，不包含粗分属性

  939.      * @return array()

  940.      */

  941.      function GetSimpleResult()

  942.      {

  943.         $rearr = array();

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

  945.         {

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

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

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

  949.         }

  950.         return $rearr;

  951.      }

  952.      

  953.     /**

  954.      * 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）

  955.      * @return array()

  956.      */

  957.      function GetSimpleResultAll()

  958.      {

  959.         $rearr = array();

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

  961.         {

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

  963.             if( $w != ' ' )

  964.             {

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

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

  967.             }

  968.         }

  969.         return $rearr;

  970.      }

  971.      

  972.     /**

  973.      * 获取索引hash数组

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

  975.      */

  976.      function GetFinallyIndex()

  977.      {

  978.         $rearr = array();

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

  980.         {

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

  982.             {

  983.                 continue;

  984.             }

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

  986.             if( $w == ' ' )

  987.             {

  988.                 continue;

  989.             }

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

  991.             {

  992.                  $rearr[$w]++;

  993.             }

  994.             else

  995.             {

  996.                  $rearr[$w] = 1;

  997.             }

  998.         }

  999.         return $rearr;

  1000.      }

  1001.      

  1002.     /**

  1003.      * 获得保存目标编码

  1004.      * @return int

  1005.      */

  1006.      function _source_result_charset()

  1007.      {

  1008.         if( preg_match("/^utf/", $this->targetCharSet) ) {

  1009.            $rs = 1;

  1010.         }

  1011.         else if( preg_match("/^gb/", $this->targetCharSet) ) {

  1012.            $rs = 2;

  1013.         }

  1014.         else if( preg_match("/^big/", $this->targetCharSet) ) {

  1015.            $rs = 3;

  1016.         }

  1017.         else {

  1018.             $rs = 4;

  1019.         }

  1020.         return $rs;

  1021.      }

  1022.      

  1023.      /**

  1024.      * 编译词典

  1025.      * @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>

  1026.      * 注意, 需要PHP开放足够的内存才能完成操作

  1027.      * @return void

  1028.      */

  1029.      function MakeDict( $source_file, $target_file='' )

  1030.      {

  1031.         $target_file = ($target_file=='' ? $this->mainDicFile : $target_file);

  1032.         $allk = array();

  1033.         $fp = fopen($source_file, 'r');

  1034.         while( $line = fgets($fp, 512) )

  1035.         {

  1036.             if( $line[0]=='@' ) continue;

  1037.             list($w, $r, $a) = explode(',', $line);

  1038.             $a = trim( $a );

  1039.             $w = iconv('utf-8', UCS2, $w);

  1040.             $k = $this->_get_index( $w );

  1041.             if( isset($allk[ $k ]) )

  1042.                 $allk[ $k ][ $w ] = array($r, $a);

  1043.             else

  1044.                 $allk[ $k ][ $w ] = array($r, $a);

  1045.         }

  1046.         fclose( $fp );

  1047.         $fp = fopen($target_file, 'w');

  1048.         $heade_rarr = array();

  1049.         $alldat = '';

  1050.         $start_pos = $this->mask_value * 8;

  1051.         foreach( $allk as $k => $v )

  1052.         {

  1053.             $dat  = serialize( $v );

  1054.             $dlen = strlen($dat);

  1055.             $alldat .= $dat;

  1056.         

  1057.             $heade_rarr[ $k ][0] = $start_pos;

  1058.             $heade_rarr[ $k ][1] = $dlen;

  1059.             $heade_rarr[ $k ][2] = count( $v );

  1060.         

  1061.             $start_pos += $dlen;

  1062.         }

  1063.         unset( $allk );

  1064.         for($i=0; $i < $this->mask_value; $i++)

  1065.         {

  1066.             if( !isset($heade_rarr[$i]) )

  1067.             {

  1068.                 $heade_rarr[$i] = array(0, 0, 0);

  1069.             }

  1070.             fwrite($fp, pack("Inn", $heade_rarr[$i][0], $heade_rarr[$i][1], $heade_rarr[$i][2]));

  1071.         }

  1072.         fwrite( $fp, $alldat);

  1073.         fclose( $fp );

  1074.      }

  1075.      

  1076.      /**

  1077.      * 导出词典的词条

  1078.      * @parem $targetfile 保存位置

  1079.      * @return void

  1080.      */

  1081.      function ExportDict( $targetfile )

  1082.      {

  1083.         if( !$this->mainDicHand )

  1084.         {

  1085.             $this->mainDicHand = fopen($this->mainDicFile, 'rw');

  1086.         }

  1087.         $fp = fopen($targetfile, 'w');

  1088.         for($i=0; $i <= $this->mask_value; $i++)

  1089.         {

  1090.             $move_pos = $i * 8;

  1091.             fseek($this->mainDicHand, $move_pos, SEEK_SET);

  1092.             $dat = fread($this->mainDicHand, 8);

  1093.             $arr = unpack('I1s/n1l/n1c', $dat);

  1094.             if( $arr['l'] == 0 )

  1095.             {

  1096.                 continue;

  1097.             }

  1098.             fseek($this->mainDicHand, $arr['s'], SEEK_SET);

  1099.             $data = @unserialize(fread($this->mainDicHand, $arr['l']));

  1100.             if( !is_array($data) ) continue;

  1101.             foreach($data as $k => $v)

  1102.             {

  1103.                 $w = iconv(UCS2, 'utf-8', $k);

  1104.                 fwrite($fp, "{$w},{$v[0]},{$v[1]}\n");

  1105.             }

  1106.         }

  1107.         fclose( $fp );

  1108.         return TRUE;

  1109.      }

  1110.      

  1111. 	function InportDict( $targetfile )

  1112.      {

  1113.      	if(!ini_set('memory_limit', '128M'))

  1114. 			exit('设置内存错误，请到dede官网下载解压版的base_dic_full.dic!');

  1115.      	require_once(DEDEINC.'/zip.class.php');

  1116.      	$zip = new zip();

  1117.      	//echo $targetfile;

  1118.      	$unpackagefile = array_keys($zip->Extract($targetfile,DEDEINC.'/data/'));

  1119.      	//exit();

  1120.      	$this->MakeDict(DEDEINC.'/data/'.$unpackagefile[0]);

  1121.      	unlink(DEDEINC.'/data/'.$unpackagefile[0]);

  1122.      	return true;

  1123.      }

  1124. }