wcfsetup/install/files/lib/system/search/AbstractSearchEngine.class.php

   1 <?php
   2 namespace wcf\system\search;
   3 use wcf\system\database\util\PreparedStatementConditionBuilder;
   4 use wcf\system\SingletonFactory;
   5 use wcf\util\StringUtil;
   6
   7 /**
   8  * Default implementation for search engines, this class should be extended by
   9  * all search engines to preserve compatibility in case of interface changes.
  10  *
  11  * @author      Alexander Ebert
  12  * @copyright   2001-2019 WoltLab GmbH
  13  * @license     GNU Lesser General Public License <http://opensource.org/licenses/lgpl-license.php>
  14  * @package     WoltLabSuite\Core\System\Search
  15  */
  16 abstract class AbstractSearchEngine extends SingletonFactory implements ISearchEngine {
  17         /**
  18          * class name for preferred condition builder
  19          * @var string
  20          */
  21         protected $conditionBuilderClassName = PreparedStatementConditionBuilder::class;
  22
  23         /**
  24          * list of engine-specific special characters
  25          * @var string[]
  26          */
  27         protected $specialCharacters = [];
  28
  29         /**
  30          * @inheritDoc
  31          */
  32         public function getConditionBuilderClassName() {
  33                 return $this->conditionBuilderClassName;
  34         }
  35
  36         /**
  37          * Manipulates the search term (< and > used as quotation marks):
  38          *
  39          * - <test foo> becomes <+test* +foo*>
  40          * - <test -foo bar> becomes <+test* -foo* +bar*>
  41          * - <test "foo bar"> becomes <+test* +"foo bar">
  42          *
  43          * @see http://dev.mysql.com/doc/refman/5.5/en/fulltext-boolean.html
  44          *
  45          * @param       string          $query
  46          * @return      string
  47          */
  48         protected function parseSearchQuery($query) {
  49                 $query = StringUtil::trim($query);
  50
  51                 // expand search terms with a * unless they're encapsulated with quotes
  52                 $inQuotes = false;
  53                 $previousChar = $tmp = '';
  54                 $controlCharacterOrSpace = false;
  55                 $chars = ['+', '-', '*'];
  56                 $ftMinWordLen = $this->getFulltextMinimumWordLength();
  57                 for ($i = 0, $length = mb_strlen($query); $i < $length; $i++) {
  58                         $char = mb_substr($query, $i, 1);
  59
  60                         if ($inQuotes) {
  61                                 if ($char == '"') {
  62                                         $inQuotes = false;
  63                                 }
  64                         }
  65                         else {
  66                                 if ($char == '"') {
  67                                         $inQuotes = true;
  68                                 }
  69                                 else {
  70                                         if ($char == ' ' && !$controlCharacterOrSpace) {
  71                                                 $controlCharacterOrSpace = true;
  72                                                 $tmp .= '*';
  73                                         }
  74                                         else if (in_array($char, $chars)) {
  75                                                 $controlCharacterOrSpace = true;
  76                                         }
  77                                         else {
  78                                                 $controlCharacterOrSpace = false;
  79                                         }
  80                                 }
  81                         }
  82
  83                         /*
  84                          * prepend a plus sign (logical AND) if ALL these conditions are given:
  85                          *
  86                          * 1) previous character:
  87                          *   - is empty (start of string)
  88                          *   - is a space (MySQL uses spaces to separate words)
  89                          *
  90                          * 2) not within quotation marks
  91                          *
  92                          * 3) current char:
  93                          *   - is NOT +, - or *
  94                          */
  95                         if (($previousChar == '' || $previousChar == ' ') && !$inQuotes && !in_array($char, $chars)) {
  96                                 // check if the term is shorter than the minimum fulltext word length
  97                                 if ($i + $ftMinWordLen <= $length) {
  98                                         $term = '';// $char;
  99                                         for ($j = $i, $innerLength = $ftMinWordLen + $i; $j < $innerLength; $j++) {
 100                                                 $currentChar = mb_substr($query, $j, 1);
 101                                                 if ($currentChar == '"' || $currentChar == ' ' || in_array($currentChar, $chars)) {
 102                                                         break;
 103                                                 }
 104
 105                                                 $term .= $currentChar;
 106                                         }
 107
 108                                         if (mb_strlen($term) == $ftMinWordLen) {
 109                                                 $tmp .= '+';
 110                                         }
 111                                 }
 112                         }
 113
 114                         $tmp .= $char;
 115                         $previousChar = $char;
 116                 }
 117
 118                 // handle last char
 119                 if (!$inQuotes && !$controlCharacterOrSpace) {
 120                         $tmp .= '*';
 121                 }
 122
 123                 return $tmp;
 124         }
 125
 126         /**
 127          * Returns minimum word length for fulltext indices.
 128          *
 129          * @return      integer
 130          */
 131         abstract protected function getFulltextMinimumWordLength();
 132
 133         /**
 134          * @inheritDoc
 135          */
 136         public function removeSpecialCharacters($string) {
 137                 if (!empty($this->specialCharacters)) {
 138                         return str_replace($this->specialCharacters, '', $string);
 139                 }
 140
 141                 return $string;
 142         }
 143 }