euc/nucleus/libs/SEARCH.php

   1 <?php\r
   2 \r
   3 /**\r
   4   * Nucleus: PHP/MySQL Weblog CMS (http://nucleuscms.org/) \r
   5   * Copyright (C) 2003-2005 The Nucleus Group\r
   6   *\r
   7   * This program is free software; you can redistribute it and/or\r
   8   * modify it under the terms of the GNU General Public License\r
   9   * as published by the Free Software Foundation; either version 2\r
  10   * of the License, or (at your option) any later version.\r
  11   * (see nucleus/documentation/index.html#license for more info)\r
  12   *\r
  13   * SEARCH(querystring) offers different functionality to create an\r
  14   * SQL query to find certain items. (and comments)\r
  15   *\r
  16   * based on code by David Altherr:\r
  17   * http://www.evolt.org/article/Boolean_Fulltext_Searching_with_PHP_and_MySQL/18/15665/\r
  18   * http://davidaltherr.net/web/php_functions/boolean/funcs.mysql.boolean.txt\r
  19   * \r
  20   *\r
  21   * $Id: SEARCH.php,v 1.5 2005-03-16 08:10:35 kimitake Exp $\r
  22   $ $NucleusJP: SEARCH.php,v 1.4 2005/03/12 06:19:05 kimitake Exp $\r
  23   */\r
  24 \r
  25 \r
  26 class SEARCH {\r
  27         \r
  28         var $querystring;\r
  29         var $marked;\r
  30         var $inclusive;\r
  31         var $blogs;\r
  32 \r
  33 \r
  34     function SEARCH($text) {\r
  35         global $blogid;\r
  36 \r
  37 //       $text = preg_replace ("/[<,>,=,?,!,#,^,(,),[,\],:,;,\\\,%]/","",$text);\r
  38 \r
  39      /* * * for jp * * * * * * * * * * */\r
  40         $text = $this->zenspace_replace($text);\r
  41         $text = preg_replace ("/[<>=?!#^()[\]:;\\%]/","",$text);\r
  42 \r
  43         $this->ascii = '[\x00-\x7F]';\r
  44         $this->two = '[\x8E\xA1-\xFE][\xA1-\xFE]';\r
  45         $this->three = '\x8F[\xA1-\xFE][\xA1-\xFE]';\r
  46 \r
  47         $this->jpmarked      = $this->boolean_mark_atoms_jp($text);\r
  48      /* * * * * * * * * * * * * * * * */\r
  49 \r
  50         $this->querystring      = $text;\r
  51 //        $this->marked         = $this->boolean_mark_atoms($text);\r
  52         $this->inclusive        = $this->boolean_inclusive_atoms($text);\r
  53         $this->blogs            = array();\r
  54 \r
  55         // get all public searchable blogs, no matter what, include the current blog allways.\r
  56                 $res = sql_query('SELECT bnumber FROM '.sql_table('blog').' WHERE bincludesearch=1 ');\r
  57                 while ($obj = mysql_fetch_object($res)) \r
  58                     $this->blogs[] = intval($obj->bnumber);\r
  59         }\r
  60 /***********************************************\r
  61   zenkaku space to space\r
  62 ***********************************************/\r
  63 \r
  64     function zenspace_replace($text){\r
  65         $ta = unpack("C*",$text);\r
  66         $len = count($ta);\r
  67         $temp = '';\r
  68         for($i=1; $i<=$len; $i++){\r
  69                 if($ta[$i]>= 0x8e){ // ja\r
  70                         if($ta[$i]>0xa0 || $ta[$i]<0x8f){ // 2byte\r
  71                                 $char = chr($ta[$i]).chr($ta[$i+1]);\r
  72                                 if($char == "\xA1\xA1") $char = "\x20";\r
  73                                 $temp .= $char;\r
  74                                 $i ++;\r
  75                         }else{ // 3byte\r
  76                                 $temp .= chr($ta[$i]).chr($ta[$i+1]).chr($ta[$i+2]);\r
  77                                 $i += 2;\r
  78                         }\r
  79                 }else{ //ascii\r
  80                         $temp .= chr($ta[$i]);\r
  81                 }\r
  82         }\r
  83       return $temp;\r
  84    }\r
  85 /***********************************************/\r
  86 \r
  87     function  boolean_sql_select($match){\r
  88 //        $string = $this->inclusive;\r
  89 //        if (strlen($string) > 0) {\r
  90            /* build sql for determining score for each record */\r
  91 /*\r
  92                preg_match_all(\r
  93                                    "([A-Za-z0-9]{1,}[A-Za-z0-9\-\.\_]{0,})",\r
  94                                $string,\r
  95                                $result);\r
  96            $result = $result[0];\r
  97                  for($cth=0;$cth<count($result);$cth++){\r
  98              if(strlen($result[$cth])>=4){\r
  99                    $stringsum_long .=  " $result[$cth] ";\r
 100                      }else{\r
 101                        $stringsum_a[] = ' '.$this->boolean_sql_select_short($result[$cth],$match).' ';\r
 102                      }\r
 103                  }\r
 104              if(strlen($stringsum_long)>0){\r
 105                                 $stringsum_long = addslashes($stringsum_long);\r
 106                         $stringsum_a[] = " match ($match) against ('$stringsum_long') ";\r
 107              }\r
 108                  $stringsum .= implode("+",$stringsum_a);\r
 109              return $stringsum;\r
 110             }\r
 111 */\r
 112     }\r
 113 \r
 114     \r
 115 \r
 116     function boolean_inclusive_atoms($string){\r
 117         $result=trim($string);\r
 118         $result=preg_replace("/([[:space:]]{2,})/",' ',$result);\r
 119 \r
 120         /* convert normal boolean operators to shortened syntax */\r
 121         $result=eregi_replace(' not ',' -',$result);\r
 122         $result=eregi_replace(' and ',' ',$result);\r
 123         $result=eregi_replace(' or ',',',$result);\r
 124 \r
 125         /* drop unnecessary spaces */\r
 126         $result=str_replace(' ,',',',$result);\r
 127         $result=str_replace(', ',',',$result);\r
 128         $result=str_replace('- ','-',$result);\r
 129         $result=str_replace('+','',$result);\r
 130 \r
 131         /* strip exlusive atoms */\r
 132         $result=preg_replace(\r
 133 //              "(\-\([A-Za-z0-9]{1,}[A-Za-z0-9\-\.\_\,]{0,}\))",\r
 134                 "(\-\(([A-Za-z0-9]|$this->two|$this->three){1,}([A-Za-z0-9\-\.\_\,]|$this->two|$this->three){0,}\))",\r
 135                 '',\r
 136                 $result);\r
 137 \r
 138         $result=preg_replace(\r
 139 //              "(\-[A-Za-z0-9]{1,}[A-Za-z0-9\-\.\_]{0,})",\r
 140                 "(\-([A-Za-z0-9]|$this->two|$this->three){1,}([A-Za-z0-9\-\.\_\,]|$this->two|$this->three){0,})",\r
 141                 '',\r
 142                 $result);\r
 143 \r
 144         $result=str_replace('(',' ',$result);\r
 145         $result=str_replace(')',' ',$result);\r
 146         $result=str_replace(',',' ',$result);\r
 147 \r
 148         return $result;\r
 149     }\r
 150 \r
 151     function boolean_sql_where($match){\r
 152 /*\r
 153         $result = $this->marked;\r
 154         $result = preg_replace(\r
 155                 "/foo\[\(\'([^\)]{4,})\'\)\]bar/e",\r
 156                 " 'match ('.\$match.') against (\''.\$this->copyvalue(\"$1\").'\') > 0 ' ",\r
 157                 $result);\r
 158 \r
 159         $result = preg_replace(                 \r
 160             "/foo\[\(\'([^\)]{1,3})\'\)\]bar/e",\r
 161             " '('.\$this->boolean_sql_where_short(\"$1\",\"$match\").')' ",             \r
 162             $result);\r
 163 */\r
 164       $result = $this->jpmarked; /* for jp */\r
 165         $result = $this->boolean_sql_where_jp_short($result,$match);/* for jp */\r
 166         return $result;\r
 167     }\r
 168 \r
 169     // there must be a simple way to simply copy a value with backslashes in it through\r
 170     // the preg_replace, but I cannot currently find it (karma 2003-12-30)\r
 171     function copyvalue($foo) {\r
 172         return $foo;\r
 173     }\r
 174 \r
 175 /***********************************************\r
 176   Make "WHERE" (jp)\r
 177 ***********************************************/\r
 178 \r
 179     function boolean_mark_atoms_jp($string){\r
 180 \r
 181         $result=trim($string);\r
 182         $result=preg_replace("/([[:space:]]{2,})/",' ',$result);\r
 183         \r
 184         /* convert normal boolean operators to shortened syntax */\r
 185         $result=eregi_replace(' not ',' -',$result);\r
 186         $result=eregi_replace(' and ',' ',$result);\r
 187         $result=eregi_replace(' or ',',',$result);\r
 188 \r
 189         /* strip excessive whitespace */\r
 190         $result=str_replace(', ',',',$result);\r
 191         $result=str_replace(' ,',',',$result);\r
 192         $result=str_replace('- ','-',$result);\r
 193         $result=str_replace('+','',$result);\r
 194         \r
 195         $result=str_replace(',',' ,',$result);\r
 196 \r
 197         return $result;\r
 198     }\r
 199     \r
 200 \r
 201     function boolean_sql_where_jp_short($string,$match){\r
 202         $match_a = explode(',',$match);\r
 203                         $key_a = explode(' ',$string);\r
 204                         \r
 205         for($ith=0;$ith<count($match_a);$ith++){\r
 206                 $temp_a[$ith] = "(i.$match_a[$ith] LIKE '%" . addslashes($key_a[0]) . "%') ";\r
 207         }\r
 208         $like = '('.implode(' or ',$temp_a).')';\r
 209 \r
 210                         for($kn=1; $kn<count($key_a); $kn++){\r
 211                 if(substr($key_a[$kn],0,1) == ","){\r
 212                         for($ith=0;$ith<count($match_a);$ith++){\r
 213                                 $temp_a[$ith] = " (i.$match_a[$ith] LIKE '%" . addslashes(substr($key_a[$kn],1)) . "%') ";\r
 214                         }\r
 215                         $like .=' OR ('. implode(' or ',$temp_a).')';\r
 216                 }elseif(substr($key_a[$kn],0,1) != '-'){\r
 217                         for($ith=0;$ith<count($match_a);$ith++){\r
 218                                 $temp_a[$ith] = " (i.$match_a[$ith] LIKE '%" . addslashes($key_a[$kn]) . "%') ";\r
 219                         }\r
 220                         $like .=' AND ('. implode(' or ',$temp_a).')';\r
 221                 }else{\r
 222                         for($ith=0;$ith<count($match_a);$ith++){\r
 223                                 $temp_a[$ith] = " NOT(i.$match_a[$ith] LIKE '%" . addslashes(substr($key_a[$kn],1)) . "%') ";\r
 224                         }\r
 225                         $like .=' AND ('. implode(' and ',$temp_a).')';\r
 226                 }\r
 227         }\r
 228         \r
 229         $like = '('.$like.')';\r
 230         return $like;\r
 231     }\r
 232 \r
 233 /***********************************************/\r
 234 \r
 235 \r
 236 /*\r
 237     function boolean_mark_atoms($string){\r
 238         $result=trim($string);\r
 239         $result=preg_replace("/([[:space:]]{2,})/",' ',$result);\r
 240 \r
 241         //convert normal boolean operators to shortened syntax\r
 242         $result=eregi_replace(' not ',' -',$result);\r
 243         $result=eregi_replace(' and ',' ',$result);\r
 244         $result=eregi_replace(' or ',',',$result);\r
 245 \r
 246 \r
 247         //strip excessive whitespace\r
 248         $result=str_replace('( ','(',$result);\r
 249         $result=str_replace(' )',')',$result);\r
 250         $result=str_replace(', ',',',$result);\r
 251         $result=str_replace(' ,',',',$result);\r
 252         $result=str_replace('- ','-',$result);\r
 253         $result=str_replace('+','',$result);\r
 254 \r
 255         // remove double spaces (we might have introduced some new ones above)\r
 256         $result=trim($result);\r
 257         $result=preg_replace("/([[:space:]]{2,})/",' ',$result);\r
 258 \r
 259         // apply arbitrary function to all 'word' atoms\r
 260 \r
 261         $result_a = explode(" ",$result);\r
 262         for($word=0;$word<count($result_a);$word++){\r
 263             $result_a[$word] = "foo[('".$result_a[$word]."')]bar";\r
 264         }\r
 265         $result = implode(" ",$result_a);\r
 266         \r
 267         // dispatch ' ' to ' AND '\r
 268         $result=str_replace(' ',' AND ',$result);\r
 269 \r
 270         // dispatch ',' to ' OR '\r
 271         $result=str_replace(',',' OR ',$result);\r
 272 \r
 273         // dispatch '-' to ' NOT '\r
 274         $result=str_replace(' -',' NOT ',$result);\r
 275         return $result;\r
 276     }\r
 277     \r
 278     function boolean_sql_where_short($string,$match){\r
 279         $match_a = explode(',',$match);\r
 280         for($ith=0;$ith<count($match_a);$ith++){\r
 281                 $like_a[$ith] = " $match_a[$ith] LIKE '% $string %' ";\r
 282         }\r
 283         $like = implode(" OR ",$like_a);\r
 284 \r
 285         return $like;\r
 286     }\r
 287     function boolean_sql_select_short($string,$match){\r
 288         $match_a = explode(',',$match);\r
 289         $score_unit_weight = .2;\r
 290         for($ith=0;$ith<count($match_a);$ith++){\r
 291             $score_a[$ith] =\r
 292                            " $score_unit_weight*(\r
 293                            LENGTH(" . addslashes($match_a[$ith]) . ") -\r
 294                            LENGTH(REPLACE(LOWER(" . addslashes($match_a[$ith]) . "),LOWER('" . addslashes($string) . "'),'')))\r
 295                                        /LENGTH('" . addslashes($string) . "') ";\r
 296         }\r
 297             $score = implode(" + ",$score_a);\r
 298 \r
 299         return $score;\r
 300     }\r
 301 */\r
 302 \r
 303 }\r
 304 ?>\r