search_function.php

完美的在线教育系统

<?php
/*
----------------------------------------------------------------------------------
PhpDig Version 1.8.x - See the config file for the full version number.
This program is provided WITHOUT warranty under the GNU/GPL license.
See the LICENSE file for more information about the GNU/GPL license.
Contributors are listed in the CREDITS and CHANGELOG files in this package.
Developer from inception to and including PhpDig v.1.6.2: Antoine Bajolet
Developer from PhpDig v.1.6.3 to and including current version: Charter
Copyright (C) 2001 - 2003, Antoine Bajolet, http://www.toiletoine.net/
Copyright (C) 2003 - current, Charter, http://www.phpdig.net/
Contributors hold Copyright (C) to their code submissions.
Do NOT edit or remove this copyright or licence information upon redistribution.
If you modify code and redistribute, you may ADD your copyright to this notice.
----------------------------------------------------------------------------------
*/

//===============================================
// do the search and display the results
// can be called in any page
function phpdigSearch($id_connect, $query_string, $option='start', $refine=0,
                       $refine_url='', $lim_start=0, $limite=10, $browse=0,
                       $site=0, $path='', $relative_script_path = '.', $template='', $adlog_flag=0, $rssdf='', $template_demo='')
{

// check input

// $id_connect set in connect.php file
// $query_string cleaned in $query_to_parse in search_function.php file
if (($option != "start") && ($option != "any") && ($option != "exact")) { $option = SEARCH_DEFAULT_MODE; }
if (($refine != 0) && ($refine != 1)) { $refine = 0; }
// $refine_url set in search_function.php file
// $lim_start set in search_function.php file
settype($limite,'integer');
if (($limite != 10) && ($limite != 30) && ($limite != 100)) { $limite = SEARCH_DEFAULT_LIMIT; }
settype($limit_start,'integer');
if (isset($limit_start)) { $limit_start = $limite * floor($limit_start / $limite); }
if (($browse != 0) && ($browse != 1)) { $browse = 0; }
if (eregi("^[0-9]+[,]",$site)) { $tempbust = explode(",",$site); $site = $tempbust[0]; $path = $tempbust[1]; }
settype($site,'integer'); // now set to integer
settype($path,'string');  // make sure set to string
if (!get_magic_quotes_gpc()) { $my_path = addslashes($path); }
else { $my_path = addslashes(stripslashes($path)); $path = stripslashes($path); }
if (empty($site) && empty($path)) { $refine = 0; } else { $refine = 1; }
if ($path == "-###-") { $site = 0; $path = ""; $refine = 0; }
// $relative_script_path set in search.php file
// $template set in config.php file
// $adlog_flag set in search.php file
// $rssdf set in search.php file
// $template_demo set in config.php

$timer = new phpdigTimer('html');
$timer->start('All');

// init variables
global $phpdig_words_chars;
settype($maxweight,'integer');
$ignore = '';
$ignore_common = '';
$ignore_message = '';
$ignore_commess = '';
$wheresite = '';
$wherepath = '';
$table_results = '';
$final_result = '';
$search_time = 0;
$strings = '';
$num_tot = 0;
$leven_final = "";
$exclude = array();
$nav_bar = '';
$pages_bar = '';

$mtime = explode(' ',microtime());
$start_time = $mtime[0]+$mtime[1];

$timer->start('All backend');
$timer->start('parsing strings');

if (!$option) {
     $option = SEARCH_DEFAULT_MODE;
}
if (!in_array($option,array('start','any','exact'))) {
     return 0;
}
// the query was filled
if ($query_string) {

$common_words = phpdigComWords("$relative_script_path/includes/common_words.txt");

$like_start = array( "start" => "", // is empty
                     "any" => "", // was a percent
                     "exact" => "" // is empty
                     );
$like_end = array( "start" => "%", // is a percent
                     "any" => "%", // is a percent
                     "exact" => "%" // was empty
                     );
$like_operator = array( "start" => "like", // is a like
                     "any" => "like", // is a like
                     "exact" => "like" // was an =
                     );

if ($refine) {
     $wheresite = "AND spider.site_id = $site ";
     if (($path) && (strlen($path) > 0)) {
          $wherepath = "AND spider.path like '$my_path' ";
     }
     $refine_url = "&amp;refine=1&amp;site=$site&amp;path=".urlencode($path);
}
else {
     $refine_url = "";
}

settype ($lim_start,"integer");
if ($lim_start < 0) {
     $lim_start = 0;
}

$n_words = count(explode(" ",$query_string));

$ncrit = 0;
$tin = "0";

if (!get_magic_quotes_gpc()) {
    $query_to_parse = addslashes($query_string);
}
else {
    $query_to_parse = $query_string;
}
$my_query_string_link = stripslashes($query_to_parse);

$query_to_parse = str_replace('_','\_',$query_to_parse); // avoid '_' in the query
$query_to_parse = str_replace('%','\%',$query_to_parse); // avoid '%' in the query
$query_to_parse = str_replace('\"',' ',$query_to_parse); // avoid '"' in the query

$query_to_parse = phpdigStripAccents(strtolower($query_to_parse)); //made all lowercase

// RH query_chars = "[^".$phpdig_words_chars[PHPDIG_ENCODING]." \'.\_~@#$:&\%/;,=-]+"; // epure chars \'._~@#$:&%/;,=-
$what_query_chars = "[^".$phpdig_words_chars[PHPDIG_ENCODING]." \'._~@#$&%/=-]+"; // epure chars \'._~@#$&%/=-

if (eregi($what_query_chars,$query_to_parse)) {
	$query_to_parse = eregi_replace($what_query_chars," ",$query_to_parse);
}

$query_to_parse = ereg_replace('(['.$phpdig_words_chars[PHPDIG_ENCODING].'])[\'.\_~@#$:&\%/;,=-]+($|[[:space:]]$|[[:space:]]['.$phpdig_words_chars[PHPDIG_ENCODING].'])','\1 \2',$query_to_parse);

$query_to_parse = trim(ereg_replace(" +"," ",$query_to_parse)); // no more than 1 blank

$query_for_strings = $query_to_parse;
$query_for_phrase = $query_to_parse;
$test_short = $query_to_parse;
$query_to_parse2 = explode(" ",$query_to_parse);
usort($query_to_parse2, "phpdigByLength");
$query_to_parse = implode(" ",$query_to_parse2);
if (isset($query_to_parse2)) { unset($query_to_parse2); }

if (SMALL_WORDS_SIZE >= 1) {
$ignore_short_flag = 0;
$test_short_counter = 0;
$test_short2 = explode(" ",$test_short);
for ($i=0; $i<count($test_short2); $i++) {
    $test_short2[$i] = trim($test_short2[$i]);
}
$test_short2 = array_unique($test_short2);
sort($test_short2);
$test_short3 = array();
for ($i=0; $i<count($test_short2); $i++) {
  if ((strlen($test_short2[$i]) <= SMALL_WORDS_SIZE) && (strlen($test_short2[$i]) > 0)) {
    $test_short2[$i].=" ";
    $test_short_counter++;
    $test_short3[] = $test_short2[$i];
  }
}
$test_short = implode(" ",$test_short3);
if (isset($test_short2)) { unset($test_short2); }
if (isset($test_short3)) { unset($test_short3); }

  $regs = array(); // for use with ereg()
  while (ereg('( [^ ]{1,'.SMALL_WORDS_SIZE.'} )|( [^ ]{1,'.SMALL_WORDS_SIZE.'})$|^([^ ]{1,'.SMALL_WORDS_SIZE.'} )',$test_short,$regs)) {
     for ($n=1; $n<=3; $n++) {
        if (($regs[$n]) || ($regs[$n] == 0)) {
            $ignore_short_flag++;
            if (!eregi("\"".trim(stripslashes($regs[$n]))."\", ",$ignore)) {
              $ignore .= "\"".trim(stripslashes($regs[$n]))."\", ";
            }
            $test_short = trim(str_replace($regs[$n],"",$test_short));
        }
     }
  }
  if (strlen($test_short) <= SMALL_WORDS_SIZE) {
    if (!eregi("\"".$test_short."\", ",$ignore)) {
      $ignore_short_flag++;
      $ignore .= "\"".stripslashes($test_short)."\", ";
    }
    $test_short = trim(str_replace($test_short,"",$test_short));
  }
}

$ignore = str_replace("\"\", ","",$ignore);

if ($option != "exact") {
  if (($ignore) && ($ignore_short_flag > 1) && ($test_short_counter > 1)) {
    $ignore_message = $ignore.' '.phpdigMsg('w_short_plur');
  }
  elseif ($ignore) {
    $ignore_message = $ignore.' '.phpdigMsg('w_short_sing');
  }
}

$ignore_common_flag = 0;
while (ereg("(-)?([^ ]{".(SMALL_WORDS_SIZE+1).",}).*",$query_for_strings,$regs)) {
        $query_for_strings = trim(str_replace($regs[2],"",$query_for_strings));
        if (!isset($common_words[stripslashes($regs[2])])) {
             if ($regs[1] == '-') {
                 $exclude[$ncrit] = $regs[2];
                 $query_for_phrase = trim(str_replace("-".$regs[2],"",$query_for_phrase));
             }
             else {
                 $strings[$ncrit] = $regs[2];
             }
             $kconds[$ncrit] = '';
             if ($option != 'any') {
                 $kconds[$ncrit] .= " AND k.twoletters = '".addslashes(substr(str_replace('\\','',$regs[2]),0,2))."' ";
             }
             $kconds[$ncrit] .= " AND k.keyword ".$like_operator[$option]." '".$like_start[$option].$regs[2].$like_end[$option]."' ";
             $ncrit++;
        }
        else {
             $ignore_common_flag++;
             $ignore_common .= "\"".stripslashes($regs[2])."\", ";
        }
}

if ($option != "exact") {
  if (($ignore_common) && ($ignore_common_flag > 1)) {
    $ignore_commess = $ignore_common.' '.phpdigMsg('w_common_plur');
  }
  elseif ($ignore_common) {
    $ignore_commess = $ignore_common.' '.phpdigMsg('w_common_sing');
  }
}

$timer->stop('parsing strings');

if ($ncrit && is_array($strings)) {
     $query = "SET OPTION SQL_BIG_SELECTS = 1";
     mysql_query($query,$id_connect);

     $my_spider2site_array = array();
     $my_sitecount_array = array();

     for ($n = 0; $n < $ncrit; $n++) {
           $timer->start('spider queries');

           $query = "SELECT spider.spider_id,sum(weight) as weight, spider.site_id
           FROM ".PHPDIG_DB_PREFIX."keywords as k,".PHPDIG_DB_PREFIX."engine as engine, ".PHPDIG_DB_PREFIX."spider as spider
           WHERE engine.key_id = k.key_id
           ".$kconds[$n]."
           AND engine.spider_id = spider.spider_id $wheresite $wherepath
           GROUP BY spider.spider_id,spider.site_id ";

           $result = mysql_query($query,$id_connect);
           $num_res_temp = mysql_num_rows($result);

           $timer->stop('spider queries');
           $timer->start('spider fills');

           if ($num_res_temp > 0) {
               if (!isset($exclude[$n])) {
               $num_res[$n] = $num_res_temp;
                    while (list($spider_id,$weight,$site_id) = mysql_fetch_array($result)) {
                         $s_weight[$n][$spider_id] = $weight;
                         $my_spider2site_array[$spider_id] = $site_id;
                         $my_sitecount_array[$site_id] = 0;
                    }
               }
               else {
               $num_exclude[$n] = $num_res_temp;
                     while (list($spider_id,$weight) = mysql_fetch_array($result)) {
                            $s_exclude[$n][$spider_id] = 1;
                     }
               mysql_free_result($result);
               }
           }
           elseif (!isset($exclude[$n])) {
                   $num_res[$n] = 0;
                   $s_weight[$n][0] = 0;
          }
          $timer->stop('spider fills');
     }

     $timer->start('reorder results');

     if ($option != "any") {
     if (is_array($num_res)) {
           asort ($num_res);
           list($id_most) = each($num_res);
           reset ($s_weight[$id_most]);
           while (list($spider_id,$weight) = each($s_weight[$id_most]))  {
                  $weight_tot = 1;
                  reset ($num_res);
                  while(list($n) = each($num_res)) {
                        settype($s_weight[$n][$spider_id],'integer');
                        $weight_tot *= sqrt($s_weight[$n][$spider_id]);
                  }
                  if ($weight_tot > 0) {
                       $final_result[$spider_id]=$weight_tot;
                  }
           }
     }
     }
     else {
     if (is_array($num_res)) {
           asort ($num_res);
           while (list($spider_id,$site_id) = each($my_spider2site_array))  {
                  $weight_tot = 0;
                  reset ($num_res);
                  while(list($n) = each($num_res)) {
                        settype($s_weight[$n][$spider_id],'integer');
                        $weight_tot += sqrt($s_weight[$n][$spider_id]);
                  }
                  if ($weight_tot > 0) {
                       $final_result[$spider_id]=$weight_tot;
                  }
           }
     }
     }

    if (isset($num_exclude) && is_array($num_exclude)) {
           while (list($id) = each($num_exclude)) {
                  while(list($spider_id) = each($s_exclude[$id])) {
                        if (isset($final_result[$spider_id])) { unset($final_result[$spider_id]); }
                  }
           }
    }

    if ($option == "exact") {
    if ((is_array($final_result)) && (count($final_result) > 0)) {
        $exact_phrase_flag = 0;
        arsort($final_result);
        reset($final_result);
        $query_for_phrase_array = explode(" ",$query_for_phrase);
        $reg_strings = str_replace('@#@',' ',phpdigPregQuotes(str_replace('\\','',implode('@#@',$query_for_phrase_array))));
        $stop_regs = "[][(){}[:blank:]=&?!&#%\$