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.
1291 Commits
3 Branches
110MB
Tree: 854fa36cf3
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '854fa36cf3'
${ noResults }
DedeV6/src/system/libraries/splitword.class.php

1034 lines
37KB
Raw Blame History 

  1. <?php

  2. if (!defined('DEDEINC')) exit('dedebiz');

  3. /**

  4.  * Unicode编码词典的php分词器

  5.  *

  6.  *  1、只适用于php5，必要函数 iconv

  7.  *  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

  8.  *  3、简单操作流程：SetSource -> StartAnalysis -> Get***Result

  9.  *  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

  10.  *

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

  12.  * @package        DedeBIZ.Libraries

  13.  * @copyright      Copyright (c) 2022 DedeBIZ.COM

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

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

  16.  */

  17. //常量定义

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

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

  20. class SplitWord

  21. {

  22.     //hash算法选项

  23.     var $mask_value = 0xFFFF;

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

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

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

  27.     //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

  28.     var $resultType = 1;

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

  30.     var $notSplitLen = 5;

  31.     //把英文单词全部转小写

  32.     var $toLower = FALSE;

  33.     //使用最大切分模式对二元词进行消岐

  34.     var $differMax = FALSE;

  35.     //尝试合并单字

  36.     var $unitWord = TRUE;

  37.     //初始化类时直接加载词典

  38.     var $loadInit = TRUE;

  39.     //使用热门词优先模式进行消岐

  40.     var $differFreq = FALSE;

  41.     //被转换为unicode的源字符串

  42.     var $sourceString = '';

  43.     //附加词典

  44.     var $addonDic = array();

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

  46.     //主词典 

  47.     var $dicStr = '';

  48.     var $mainDic = array();

  49.     var $mainDicHand = FALSE;

  50.     var $mainDicInfos = array();

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

  52.     //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）

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

  54.     var $isLoadAll = FALSE;

  55.     var $isUnpacked = FALSE;

  56.     //主词典词语最大长度 x / 2

  57.     var $dicWordMax = 14;

  58.     //粗分后的数组（通常是截取句子等用途）

  59.     var $simpleResult = array();

  60.     //最终结果(用空格分开的词汇列表)

  61.     var $finallyResult = array();

  62.     //是否已经载入词典

  63.     var $isLoadDic = FALSE;

  64.     //系统识别或合并的新词

  65.     var $newWords = array();

  66.     var $foundWordStr = '';

  67.     //词库载入时间

  68.     var $loadTime = 0;

  69.     /**

  70.      * 构造函数

  71.      * @param $source_charset

  72.      * @param $target_charset

  73.      * @param $load_alldic 

  74.      * @param $source

  75.      *

  76.      * @return void

  77.      */

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

  79.     {

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

  81.         $this->isLoadAll = $load_all;

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

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

  84.     }

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

  86.     {

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

  88.     }

  89.     /**

  90.     * 析构函数

  91.     */

  92.     function __destruct()

  93.     {

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

  95.         {

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

  97.         }

  98.     }

  99.     /**

  100.      * 根据字符串计算key索引

  101.      * @param $key

  102.      * @return short int

  103.      */

  104.     function _get_index( $key )

  105.     {

  106.         $l = strlen($key);

  107.         $h = 0x238f13af;

  108.         while ($l--)

  109.         {

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

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

  112.             $h &= 0x7fffffff;

  113.         }

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

  115.     }

  116.     /**

  117.      * 从文件获得词

  118.      * @param $key

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

  120.      * @return short int

  121.      */

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

  123.     {

  124.         if ( !$this->mainDicHand )

  125.         {

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

  127.         }

  128.         $p = 0;

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

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

  131.         {

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

  133.         } else {

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

  135.             $move_pos = $keynum * 8;

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

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

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

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

  140.             {

  141.                 return FALSE;

  142.             }

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

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

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

  146.        }

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

  148.        {

  149.            return FALSE;

  150.        }

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

  152.     }

  153.     /**

  154.      * 设置源字符串

  155.      * @param $source

  156.      * @param $source_charset

  157.      * @param $target_charset

  158.      *

  159.      * @return bool

  160.      */

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

  162.     {

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

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

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

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

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

  168.         if ( $source != '' )

  169.         {

  170.             $rs = TRUE;

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

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

  173.             }

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

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

  176.             }

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

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

  179.             } else {

  180.                 $rs = FALSE;

  181.             }

  182.         } else {

  183.            $rs = FALSE;

  184.         }

  185.         return $rs;

  186.     }

  187.     /**

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

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

  190.      *

  191.      * @return void

  192.      */

  193.     function SetResultType( $rstype )

  194.     {

  195.         $this->resultType = $rstype;

  196.     }

  197.     /**

  198.      * 载入词典

  199.      *

  200.      * @return void

  201.      */

  202.     function LoadDict( $maindic='' )

  203.     {

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

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

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

  207.         $startt = microtime(TRUE);

  208.         //正常读取文件

  209.         $dicAddon = $this->addonDicFile;

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

  211.         {

  212.             $dicWords = $this->mainDicFile ;

  213.         } else {

  214.             $dicWords = $maindic;

  215.             $this->mainDicFile = $maindic;

  216.         }

  217.         //加载主词典（只打开）

  218.         if ($this->isUnpacked){

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

  220.         }

  221.         //载入副词典

  222.         $hw = '';

  223.         $ds = file($dicAddon);

  224.         foreach($ds as $d)

  225.         {

  226.             $d = trim($d);

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

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

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

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

  231.             } else {

  232.                 $spstr = _SP_;

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

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

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

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

  237.                 foreach($ws as $estr)

  238.                 {

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

  240.                 }

  241.             }

  242.         }

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

  244.         $this->isLoadDic = TRUE;

  245.     }

  246.     /**

  247.     * 检测某个词是否存在

  248.     */

  249.     function IsWord( $word )

  250.     {

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

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

  253.     }

  254.     /**

  255.      * 获得某个词的词性及词频信息

  256.      * @parem $word unicode编码的词

  257.      * @return void

  258.      */

  259.     function GetWordProperty($word)

  260.     {

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

  262.         {

  263.             return '/s';

  264.         }

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

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

  267.     }

  268.     /**

  269.      * 指定某词的词性信息（通常是新词）

  270.      * @parem $word unicode编码的词

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

  272.      * @return void;

  273.      */

  274.     function SetWordInfos($word, $infos)

  275.     {

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

  277.         {

  278.             return ;

  279.         }

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

  281.         {

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

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

  284.         } else {

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

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

  287.         }

  288.     }

  289.     /**

  290.      * 开始执行分析

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

  292.      * @return bool

  293.      */

  294.     function StartAnalysis($optimize=TRUE)

  295.     {

  296.         if ( !$this->isLoadDic )

  297.         {

  298.             $this->LoadDict();

  299.         }

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

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

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

  303.         $sbcArr = array();

  304.         $j = 0;

  305.         //全角与半角字符对照表

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

  307.         {

  308.             $scb = 0x20 + $j;

  309.             $j++;

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

  311.         }

  312.         //对字符串进行粗分

  313.         $onstr = '';

  314.         $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

  315.         $s = 0;

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

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

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

  319.         {

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

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

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

  323.             //ANSI字符

  324.             if ($cn < 0x80)

  325.             {

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

  327.                 {

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

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

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

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

  332.                         $s++;

  333.                         $onstr = '';

  334.                     }

  335.                     $lastc = 2;

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

  337.                 } else {

  338.                     if ( $onstr != '' )

  339.                     {

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

  341.                         if ( $lastc==2 )

  342.                         {

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

  344.                         }

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

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

  347.                         $s++;

  348.                     }

  349.                     $onstr = '';

  350.                     $lastc = 3;

  351.                     if ($cn < 31)

  352.                     {

  353.                         continue;

  354.                     } else {

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

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

  357.                         $s++;

  358.                     }

  359.                 }

  360.             }

  361.             //普通字符

  362.             else

  363.             {

  364.                 //正常文字

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

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

  367.                 {

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

  369.                     {

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

  371.                         if ( $lastc==2 )

  372.                         {

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

  374.                         }

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

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

  377.                         $s++;

  378.                         $onstr = '';

  379.                     }

  380.                     $lastc = 1;

  381.                     $onstr .= $c;

  382.                 }

  383.                 //特殊符号

  384.                 else

  385.                 {

  386.                     if ( $onstr != '' )

  387.                     {

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

  389.                         if ( $lastc==2 )

  390.                         {

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

  392.                         }

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

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

  395.                         $s++;

  396.                     }

  397.                     //检测书名

  398.                     if ( $cn == 0x300A )

  399.                     {

  400.                         $tmpw = '';

  401.                         $n = 1;

  402.                         $isok = FALSE;

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

  404.                         while(TRUE)

  405.                         {

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

  407. 							break;

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

  409.                             if ( $w == $ew )

  410.                             {

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

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

  413.                                 $s++;

  414.                         

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

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

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

  418.                                 {

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

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

  421.                                 }

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

  423.                                 $s++;

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

  425.                                 if ( $this->differMax )

  426.                                 {

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

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

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

  430.                                     $s++;

  431.                                 }

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

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

  434.                                 $s++;

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

  436.                                 $isok = TRUE;

  437.                                 $onstr = '';

  438.                                 $lastc = 5;

  439.                                 break;

  440.                             } else {

  441.                                 $n = $n+2;

  442.                                 $tmpw .= $w;

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

  444.                                 {

  445.                                     break;

  446.                                 }

  447.                             }

  448.                         }//while

  449.                         if ( !$isok )

  450.                         {

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

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

  453.                               $s++;

  454.                               $onstr = '';

  455.                             $lastc = 5;

  456.                         }

  457.                         continue;

  458.                     }

  459.                     $onstr = '';

  460.                     $lastc = 5;

  461.                     if ( $cn==0x3000 )

  462.                     {

  463.                         continue;

  464.                     } else {

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

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

  467.                         $s++;

  468.                     }

  469.                 }//2byte symbol

  470.             }//end 2byte char

  471.         }//end for

  472.         //处理分词后的结果

  473.         $this->_sort_finally_result();

  474.     }

  475.     /**

  476.      * 深入分词

  477.      * @parem $str

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

  479.      * @parem $spos   当前粗分结果游标

  480.      * @return bool

  481.      */

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

  483.     {

  484.         //中文句子

  485.         if ( $ctype==1 )

  486.         {

  487.             $slen = strlen($str);

  488.             //小于系统配置分词要求长度的句子

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

  490.             {

  491.                 $tmpstr = '';

  492.                 $lastType = 0;

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

  494.                 if ($slen < 5)

  495.                 {

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

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

  498.                       {

  499.                             $str2 = '';

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

  501.                             {

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

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

  504.                             }

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

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

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

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

  509.                             {

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

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

  512.                             }

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

  514.                             if ( $str2 != '' )

  515.                             {

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

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

  518.                             }

  519.                        } else {

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

  521.                        }

  522.                 } else {

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

  524.                 }

  525.             }

  526.             //正常长度的句子，循环进行分词处理

  527.             else {

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

  529.             }

  530.         }

  531.         //英文句子，转为小写

  532.         else {

  533.             if ( $this->toLower ) {

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

  535.             } else {

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

  537.             }

  538.         }

  539.     }

  540.     /**

  541.      * 中文的深入分词

  542.      * @parem $str

  543.      * @return void

  544.      */

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

  546.     {

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

  548.         $tmparr = array();

  549.         $hasw = 0;

  550.         //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

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

  552.         {

  553.             $tmparr[] = $str;

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

  555.             {

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

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

  558.             }

  559.             if ( !$this->differMax )

  560.             {

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

  562.                 return ;

  563.             }

  564.         }

  565.         //进行切分

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

  567.         {

  568.             //单个词

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

  570.             //是否已经到最后两个字

  571.             if ( $i <= 2 )

  572.             {

  573.                 $tmparr[] = $nc;

  574.                 $i = 0;

  575.                 break;

  576.             }

  577.             $isok = FALSE;

  578.             $i = $i + 1;

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

  580.             {

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

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

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

  584.                 {

  585.                     $i = $i - 1;

  586.                     break;

  587.                 }

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

  589.                 {

  590.                     $tmparr[] = $w;

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

  592.                     $isok = TRUE;

  593.                     break;

  594.                 }

  595.             }

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

  597.             //没适合词

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

  599.         }

  600.         $wcount = count($tmparr);

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

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

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

  604.         if ( $optimize )

  605.         {

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

  607.         }

  608.     }

  609.     /**

  610.     * 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）

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

  612.     * @return bool

  613.     */

  614.     //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

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

  616.     {

  617.         $newarr = array();

  618.         $prePos = $spos - 1;

  619.         $arlen = count($smarr);

  620.         $i = $j = 0;

  621.         //检测数量词

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

  623.         {

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

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

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

  627.               {

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

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

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

  631.                  {

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

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

  634.                  }

  635.                  $smarr[0] = '';

  636.                  $i++;

  637.               }

  638.        }

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

  640.        {

  641.             

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

  643.             {

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

  645.                 break;

  646.             }

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

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

  649.             $ischeck = FALSE;

  650.             //检测数量词

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

  652.             {

  653.                 //最大切分时保留合并前的词

  654.                 if ($this->differMax)

  655.                 {

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

  657.                         $j++;

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

  659.                         $j++;

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

  661.                         $j++;

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

  663.                         $j++;

  664.                 }

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

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

  667.                 {

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

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

  670.                 }

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

  672.             }

  673.             //检测前导词(通常是姓)

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

  675.             {

  676.                 $is_rs = FALSE;

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

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

  679.                 {

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

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

  682.                     {

  683.                          $is_rs = TRUE;

  684.                     }

  685.                 }

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

  687.                 {

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

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

  690.                     //尝试检测第三个词

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

  692.                     {

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

  694.                         $i++;

  695.                     }

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

  697.                     {

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

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

  700.                     }

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

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

  703.                     {

  704.                         $j++;

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

  706.                         $j++;

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

  708.                         $j++;

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

  710.                         $j++;

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

  712.                     }

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

  714.                 }

  715.             }

  716.             //检测后缀词(地名等)

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

  718.             {

  719.                 $is_rs = FALSE;

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

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

  722.                 {

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

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

  725.                     {

  726.                          $is_rs = TRUE;

  727.                     }

  728.                 }

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

  730.                 {

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

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

  733.                     {

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

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

  736.                     }

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

  738.                 }

  739.             }

  740.             //新词识别（暂无规则）

  741.             else if ($this->unitWord)

  742.             {

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

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

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

  746.                 {

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

  748.                     //尝试检测第三个词

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

  750.                     {

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

  752.                         $i++;

  753.                     }

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

  755.                     {

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

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

  758.                     }

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

  760.                 }

  761.             }

  762.             //不符合规则

  763.             if ( !$ischeck )

  764.             {

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

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

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

  768.                 {

  769.                     $slen = strlen($nw);

  770.                     $hasDiff = FALSE;

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

  772.                     {

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

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

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

  776.                         {

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

  778.                             $hasDiff = TRUE;

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

  780.                         }

  781.                     }

  782.                 }

  783.                 $j++;

  784.             }

  785.        }//end for

  786.        $smarr =  $newarr;

  787.     }

  788.     /**

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

  790.     * @return void

  791.     */

  792.     function _sort_finally_result()

  793.     {

  794.           $newarr = array();

  795.         $i = 0;

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

  797.         {

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

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

  800.             {

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

  802.                 {

  803.                     if (!empty($w))

  804.                     {

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

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

  807.                         $i++;

  808.                     }

  809.                 }

  810.             }

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

  812.             {

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

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

  815.                 $i++;

  816.             }

  817.         }

  818.         $this->finallyResult = $newarr;

  819.         $newarr = '';

  820.     }

  821.     /**

  822.      * 把uncode字符串转换为输出字符串

  823.      * @parem str

  824.      * return string

  825.      */

  826.     function _out_string_encoding( &$str )

  827.     {

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

  829.         if ( $rsc==1 ) {

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

  831.         }

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

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

  834.         } else {

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

  836.         }

  837.         return $rsstr;

  838.     }

  839.     /**

  840.      * 获取最终结果字符串（用空格分开后的分词结果）

  841.      * @return string

  842.      */

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

  844.     {

  845.         $rsstr = '';

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

  847.         {

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

  849.             {

  850.                 continue;

  851.             }

  852.             $m = '';

  853.             if ( $word_meanings )

  854.             {

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

  856.             }

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

  858.             if ( $w != ' ' )

  859.             {

  860.                 if ($word_meanings) {

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

  862.                 } else {

  863.                     $rsstr .= $spword.$w;

  864.                 }

  865.             }

  866.         }

  867.         return $rsstr;

  868.     }

  869.     /**

  870.      * 获取粗分结果，不包含粗分属性

  871.      * @return array()

  872.      */

  873.     function GetSimpleResult()

  874.     {

  875.         $rearr = array();

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

  877.         {

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

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

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

  881.         }

  882.         return $rearr;

  883.     }

  884.     /**

  885.      * 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）

  886.      * @return array()

  887.      */

  888.     function GetSimpleResultAll()

  889.     {

  890.         $rearr = array();

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

  892.         {

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

  894.             if ( $w != ' ' )

  895.             {

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

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

  898.             }

  899.         }

  900.         return $rearr;

  901.     }

  902.     /**

  903.      * 获取索引hash数组

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

  905.      */

  906.     function GetFinallyIndex()

  907.     {

  908.         $rearr = array();

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

  910.         {

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

  912.             {

  913.                 continue;

  914.             }

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

  916.             if ( $w == ' ' )

  917.             {

  918.                 continue;

  919.             }

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

  921.             {

  922.                  $rearr[$w]++;

  923.             } else {

  924.                  $rearr[$w] = 1;

  925.             }

  926.         }

  927.         return $rearr;

  928.     }

  929.     /**

  930.      * 获得保存目标编码

  931.      * @return int

  932.      */

  933.     function _source_result_charset()

  934.     {

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

  936.            $rs = 1;

  937.         }

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

  939.            $rs = 2;

  940.         }

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

  942.            $rs = 3;

  943.         }

  944.         else {

  945.             $rs = 4;

  946.         }

  947.         return $rs;

  948.     }

  949.      /**

  950.      * 编译词典

  951.      * @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>

  952.      * 注意, 需要PHP开放足够的内存才能完成操作

  953.      * @return void

  954.      */

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

  956.     {

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

  958.         $allk = array();

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

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

  961.         {

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

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

  964.             $a = trim( $a );

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

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

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

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

  969.             else

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

  971.         }

  972.         fclose( $fp );

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

  974.         $heade_rarr = array();

  975.         $alldat = '';

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

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

  978.         {

  979.             $dat  = serialize( $v );

  980.             $dlen = strlen($dat);

  981.             $alldat .= $dat;

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

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

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

  985.             $start_pos += $dlen;

  986.         }

  987.         unset( $allk );

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

  989.         {

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

  991.             {

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

  993.             }

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

  995.         }

  996.         fwrite( $fp, $alldat);

  997.         fclose( $fp );

  998.     } 

  999.      /**

  1000.      * 导出词典的词条

  1001.      * @parem $targetfile 保存位置

  1002.      * @return void

  1003.      */

  1004.     function ExportDict( $targetfile )

  1005.     {

  1006.         if ( !$this->mainDicHand )

  1007.         {

  1008.             $this->mainDicHand = fopen($this->mainDicFile, 'rw');

  1009.         }

  1010.         $fp = fopen($targetfile, 'w');

  1011.         for($i=0; $i <= $this->mask_value; $i++)

  1012.         {

  1013.             $move_pos = $i * 8;

  1014.             fseek($this->mainDicHand, $move_pos, SEEK_SET);

  1015.             $dat = fread($this->mainDicHand, 8);

  1016.             $arr = unpack('I1s/n1l/n1c', $dat);

  1017.             if ( $arr['l'] == 0 )

  1018.             {

  1019.                 continue;

  1020.             }

  1021.             fseek($this->mainDicHand, $arr['s'], SEEK_SET);

  1022.             $data = @unserialize(fread($this->mainDicHand, $arr['l']));

  1023.             if ( !is_array($data) ) continue;

  1024.             foreach($data as $k => $v)

  1025.             {

  1026.                 $w = iconv(UCS2, 'utf-8', $k);

  1027.                 fwrite($fp, "{$w},{$v[0]},{$v[1]}\n");

  1028.             }

  1029.         }

  1030.         fclose( $fp );

  1031.         return TRUE;

  1032.      }

  1033. }