全文檢索準確度與關鍵字句高亮(Highlight)效果(原始碼)

Artvine 发表于 2004-10-25 12:43:42

全文檢索準確度與關鍵字句高亮(Highlight)效果(繁簡通用)，已經成功。原始程式資料如下↓

hilight_wizard.php -精確關鍵字標色[適用於/gb2312/big5/英美語系/unicode環境]
/*#########################################
hilight_wizard.php
** Intensive_Hilight functions Via 水電工**
精確關鍵字標色：適用於使用gb2312、big5、英美語系、unicode 之搜尋系統
所有的語系、甚至其他論壇都能用
http://artvine.com.tw
使用方法：
require_once('hilight_wizard.php');
$srchtxt = hilight_wizard($terms,$srchtxt);
BUGS & TODO：
超連結還會落入關鍵字標色範圍，待改善
################################*/

function hilight_wizard($terms , $srchtxt){
 if(got_num($terms) && is_UniCode($srchtxt))
 $srchtxt=unicode_hilight($terms , $srchtxt);
 else
 $srchtxt = highlight($terms,$srchtxt);

return $srchtxt;
}

function unicode_hilight($terms , $srchtxt){

 $arr= arr_terms($terms);

 $av=$arr;
 $ad=$arr;

 if(is_array($av))foreach( $av AS $v )$av2.= " ".$v;
 if(is_array($ad))foreach( $ad AS $v )$ad2.= " ".$v;

 $v_ret=trim($av2);
 $d_ret=trim($ad2);

 if($v_ret):

 preg_match_all("/(&#)()({4})([;])/",$srchtxt, $uni);
 $matches=array_unique($uni);
 $list_arr=uni_fake($matches,$srchtxt);

 $fake_uni_str=strtr($srchtxt, $list_arr);
 $highlight_str=hilight_uni($v_ret, $fake_uni_str);
 $srchtxt=strtr($highlight_str, $list_arr);

 endif;

 if($d_ret) $srchtxt= hilight_uni($d_ret, $srchtxt);

 return $srchtxt;
}

function is_chinese($str){
if(preg_match_all("/([\xa1-\xf9][\x40-\xfe])/",$str,$chi)>0)
 return true;
}

function got_num($terms){
 $text=chop_terms($terms);
 foreach(explode(' ', $text) as $ret){
 if(is_numeric($ret)){
 $num= true;
 break;
 }
 }
 return $num;
}

function is_UniCode($str){
 if(preg_match_all("/([&#]{4}[;])/",$str,$uni)>0)
 return true;
}

function single_space($text) {
 $text=ereg_replace(' +', ' ', trim($text));
 return $text;
}

function chop_terms($text) {

 $search_terms=array("：","，",",",":","<",">","?","/","=","+","-"," the "," and "," an "," in "," on ", " or ");
 $text=str_replace($search_terms," ",$text);
 $text=single_space($text);

 return $text;
}

function arr_terms($text) {

 $text=chop_terms($text);
 $arr = explode(" ", $text);
 foreach($arr as $v){
 if(is_numeric($v))$av[]=$v;
 else $ad[]=chr(10+strlen($v)).':'.$v;
 }

 if(is_array($av)){
 rsort($av);
 reset($av);
 }

 if(is_array($ad)){
 rsort($ad);
 reset($ad);
 }

 for ($i=0; $i list($n,$p) = explode(':',$ad[$i]);
 $newad[]= $p;
 }

 return array($av,$newad);
}

function sort_terms($terms) {
 $arr=arr_terms(stripslashes(stripslashes(stripslashes($terms))));
 $new = array_merge($arr ,$arr);

 for ($i=0; $i $new_terms.= $new[$i].'+';
 }

 return $new_terms;
}

function highlight($terms,$srchtxt,$fcolor='yellow'){
 $terms=sort_terms($terms);
 foreach(explode('+', $terms) as $ret){
 if((is_chinese($ret)==false && strlen($ret)>3) || is_chinese($ret))
 $srchtxt = preg_replace('|('.quotemeta($ret).')|iU',
 '\\1',$srchtxt);
 }

return $srchtxt;
}

function hilight_uni($terms,$srchtxt,$fcolor='yellow')
{
 $terms=stripslashes(stripslashes(stripslashes($terms)));
 foreach(explode(' ', $terms) as $ret){
 $srchtxt = preg_replace('|('.quotemeta($ret).')|iU',
 '\\1',$srchtxt);
 }
return $srchtxt;
}

function uni_fake($matches,$str){

 $fake = array(0=>'z',1=>'y',2=>'x',3=>'w',4=>'v',5=>'u',6=>'t',7=>'s',8=>'r',9=>'q');

 foreach($matches as $data){
 $temp = strtr($data, $fake);
 $str1[$data]=$temp;
 $str2[$temp]=$data;
 }
 return array($str1,$str2);
}

?>

/*#########################################
以上存為hilight_wizard.php
#########################################
hilight_wizard.php 在Discuz!的安裝及配合修正(2.0~3.12 plus)
###################################################
1.把 hilight_wizard.php 放在資料夾 include/底下

2.打開 include/discuzcode.php

在169行附近，找到if ($highlight){

(原先的代碼程序可以通通刪除)把他改成：

 if ($highlight){
 require_once('hilight_wizard.php');
 $message = hilight_wizard($highlight,$message);
 }

/*底下就是原代碼程序
 if($highlight) {
 foreach(explode('+', $highlight) as $ret) {
 $ret = addcslashes($ret, "/()[]|.:!=<>?^\$");
 if($ret) {
 $message = preg_replace("/([ \f\r\t\n]|^)$ret([ \f\r\t\n]|$)/is", "\\1$ret\\2", $message);
 }
 }
 }
*/
/*#########################################
Discuz!精確搜尋search.php修正加強版
##############################################################################
#Discuz! 精確搜尋
#水電工 http://artvine.com.tw #
#
##############################################################################

##修正對象：search.php
///////////////////////////////////////////////////////////////////////////////////////////

##在62行附近找到 $srchtxt = trim($srchtxt);
##=>>把他改為
 $srchtxt=ereg_replace(' +', ' ', trim($srchtxt));
 $srchtxt=str_replace(' | ',' OR ', $srchtxt);
 $srchtxt=mysql_escape_string($srchtxt);

//////////////////////////////////////////////////////////
##在98行附近找到 if(preg_match("(AND|\+|&|\s)", $srchtxt) && !preg_match("(OR|\|)", $srchtxt)) {


##=>在下一行插入
 $srchtxt=preg_replace ("/(&#)()({4})([;])/", "uNc\\2\\3\\4", $srchtxt);

////////////////////////////////////////////////////////////
##在108行附近找到 $srchtxt = str_replace('*', '%', $srchtxt);

##=>在下一行插入
 $srchtxt=preg_replace ("/(uNc)()({4})([;])/", "&#\\2\\3\\4", $srchtxt);
/////////////OK search.php fixed//////the end/////////

Artvine 发表于 2004-10-29 01:52:46

<?php
/**
* This file contains the keyhighlighter class that highlight the chosen keyword in the current output buffer.
*
* @package keyhighlighter
*/

/**
* keyhighlighter class
*
* This class highlight the chosen keywords in the current output buffer
*
* @package keyhighlighter
* @author Setec Astronomy
* @version 1.0
* @abstract Highlight specific keywords.
* @copyright 2004
* @example sample.php A sample code.
* @link http://setecastronomy.stufftoread.com
*/

class keyhighlighter {

/**
* @access private
*/
var $preg_keywords = '';
/**
* @access private
*/
var $keywords = '';
/**
* @access private
*/
var $singlewords = false;
/**
* @access private
*/
var $replace_callback = null;

/**
* Main constructor
*
* This is the main constructor of keyhighlighter class. 
* It's the only public method of the class.
* @param string $keywords the keywords you want to highlight
* @param boolean $singlewords specify if it has to highlight also the single words.
* @param callback $replace_callback a custom callback for keyword highlight.
* <code>
* <?php
* require ('keyhighlighter.class.php');
*
* function my_highlighter ($matches) {
* return '' . $matches . '';
* }
*
* new keyhighlighter ('W3C', false, 'my_highlighter');
* readfile ('http://www.w3c.org/');
* ?>
* </code>
*/
// public function __construct ()
function keyhighlighter ($keywords, $singlewords = false, $replace_callback = null) {
 $this->keywords = $keywords;
 $this->singlewords = $singlewords;
 $this->replace_callback = $replace_callback;
 ob_start (array (&$this, 'highlight'));
}

/**
* @access private
*/
function replace ($replace_matches) {

 $patterns = array ();
 if ($this->singlewords) {
 $keywords = explode (' ', $this->preg_keywords);
 foreach ($keywords as $keyword) {
 $patterns[] = '/(?' . '>' . $keyword . '+)/si';
 }
 } else {
 $patterns[] = '/(?' . '>' . $this->preg_keywords . '+)/si';
 }

 $result = $replace_matches;

 foreach ($patterns as $pattern) {
 if (!is_null ($this->replace_callback)) {
 $result = preg_replace_callback ($pattern, $this->replace_callback, $result);
 } else {
 $result = preg_replace ($pattern, '\\0', $result);
 }
 }

 return $result;
}

/**
* @access private
*/
function highlight ($buffer) {
 $buffer = '>' . $buffer . '<';
 $this->preg_keywords = preg_replace ('/[^\w ]/si', '', $this->keywords);
 $buffer = preg_replace_callback ("/(\>(((?" . ">[^><]+)|(?R))*)\<)/is", array (&$this, 'replace'), $buffer);
 $buffer = substr ($buffer, 1, -1);
 return $buffer;
}
}

?>

页: [1]

湘里妹子学术网's Archiver

全文檢索準確度與關鍵字句高亮(Highlight)效果(原始碼)