[Summary view] [Print] [Text view]

   1  <?php
   2  /**
   3   * Zend Framework
   4   *
   5   * LICENSE
   6   *
   7   * This source file is subject to the new BSD license that is bundled
   8   * with this package in the file LICENSE.txt.
   9   * It is also available through the world-wide-web at this URL:
  10   * http://framework.zend.com/license/new-bsd
  11   * If you did not receive a copy of the license and are unable to
  12   * obtain it through the world-wide-web, please send an email
  13   * to license@zend.com so we can send you a copy immediately.
  14   *
  15   * @category   Zend
  16   * @package    Zend_Search_Lucene
  17   * @subpackage Search
  18   * @copyright  Copyright (c) 2005-2007 Zend Technologies USA Inc. (http://www.zend.com)
  19   * @license    http://framework.zend.com/license/new-bsd     New BSD License
  20   */
  21  
  22  
  23  /** Zend_Search_Lucene_Exception */
  24  require_once $CFG->dirroot.'/search/Zend/Search/Lucene/Exception.php';
  25  
  26  
  27  /**
  28   * @category   Zend
  29   * @package    Zend_Search_Lucene
  30   * @subpackage Search
  31   * @copyright  Copyright (c) 2005-2007 Zend Technologies USA Inc. (http://www.zend.com)
  32   * @license    http://framework.zend.com/license/new-bsd     New BSD License
  33   */
  34  class Zend_Search_Lucene_Search_QueryToken
  35  {
  36      /**
  37       * Token types.
  38       */
  39      const TT_WORD                 = 0;  // Word
  40      const TT_PHRASE               = 1;  // Phrase (one or several quoted words)
  41      const TT_FIELD                = 2;  // Field name in 'field:word', field:<phrase> or field:(<subquery>) pairs
  42      const TT_FIELD_INDICATOR      = 3;  // ':'
  43      const TT_REQUIRED             = 4;  // '+'
  44      const TT_PROHIBITED           = 5;  // '-'
  45      const TT_FUZZY_PROX_MARK      = 6;  // '~'
  46      const TT_BOOSTING_MARK        = 7;  // '^'
  47      const TT_RANGE_INCL_START     = 8;  // '['
  48      const TT_RANGE_INCL_END       = 9;  // ']'
  49      const TT_RANGE_EXCL_START     = 10; // '{'
  50      const TT_RANGE_EXCL_END       = 11; // '}'
  51      const TT_SUBQUERY_START       = 12; // '('
  52      const TT_SUBQUERY_END         = 13; // ')'
  53      const TT_AND_LEXEME           = 14; // 'AND' or 'and'
  54      const TT_OR_LEXEME            = 15; // 'OR'  or 'or'
  55      const TT_NOT_LEXEME           = 16; // 'NOT' or 'not'
  56      const TT_TO_LEXEME            = 17; // 'TO'  or 'to'
  57      const TT_NUMBER               = 18; // Number, like: 10, 0.8, .64, ....
  58  
  59  
  60      /**
  61       * Returns all possible lexeme types.
  62       * It's used for syntax analyzer state machine initialization
  63       *
  64       * @return array
  65       */
  66      public static function getTypes()
  67      {
  68          return array(   self::TT_WORD,
  69                          self::TT_PHRASE,
  70                          self::TT_FIELD,
  71                          self::TT_FIELD_INDICATOR,
  72                          self::TT_REQUIRED,
  73                          self::TT_PROHIBITED,
  74                          self::TT_FUZZY_PROX_MARK,
  75                          self::TT_BOOSTING_MARK,
  76                          self::TT_RANGE_INCL_START,
  77                          self::TT_RANGE_INCL_END,
  78                          self::TT_RANGE_EXCL_START,
  79                          self::TT_RANGE_EXCL_END,
  80                          self::TT_SUBQUERY_START,
  81                          self::TT_SUBQUERY_END,
  82                          self::TT_AND_LEXEME,
  83                          self::TT_OR_LEXEME,
  84                          self::TT_NOT_LEXEME,
  85                          self::TT_TO_LEXEME,
  86                          self::TT_NUMBER
  87                       );
  88      }
  89  
  90  
  91      /**
  92       * TokenCategories
  93       */
  94      const TC_WORD           = 0;   // Word
  95      const TC_PHRASE         = 1;   // Phrase (one or several quoted words)
  96      const TC_NUMBER         = 2;   // Nubers, which are used with syntax elements. Ex. roam~0.8
  97      const TC_SYNTAX_ELEMENT = 3;   // +  -  ( )  [ ]  { }  !  ||  && ~ ^
  98  
  99  
 100      /**
 101       * Token type.
 102       *
 103       * @var integer
 104       */
 105      public $type;
 106  
 107      /**
 108       * Token text.
 109       *
 110       * @var integer
 111       */
 112      public $text;
 113  
 114      /**
 115       * Token position within query.
 116       *
 117       * @var integer
 118       */
 119      public $position;
 120  
 121  
 122      /**
 123       * IndexReader constructor needs token type and token text as a parameters.
 124       *
 125       * @param integer $tokenCategory
 126       * @param string  $tokText
 127       * @param integer $position
 128       */
 129      public function __construct($tokenCategory, $tokenText, $position)
 130      {
 131          $this->text     = $tokenText;
 132          $this->position = $position + 1; // Start from 1
 133  
 134          switch ($tokenCategory) {
 135              case self::TC_WORD:
 136                  if (  strtolower($tokenText) == 'and') {
 137                      $this->type = self::TT_AND_LEXEME;
 138                  } else if (strtolower($tokenText) == 'or') {
 139                      $this->type = self::TT_OR_LEXEME;
 140                  } else if (strtolower($tokenText) == 'not') {
 141                      $this->type = self::TT_NOT_LEXEME;
 142                  } else if (strtolower($tokenText) == 'to') {
 143                      $this->type = self::TT_TO_LEXEME;
 144                  } else {
 145                      $this->type = self::TT_WORD;
 146                  }
 147                  break;
 148  
 149              case self::TC_PHRASE:
 150                  $this->type = self::TT_PHRASE;
 151                  break;
 152  
 153              case self::TC_NUMBER:
 154                  $this->type = self::TT_NUMBER;
 155                  break;
 156  
 157              case self::TC_SYNTAX_ELEMENT:
 158                  switch ($tokenText) {
 159                      case ':':
 160                          $this->type = self::TT_FIELD_INDICATOR;
 161                          break;
 162  
 163                      case '+':
 164                          $this->type = self::TT_REQUIRED;
 165                          break;
 166  
 167                      case '-':
 168                          $this->type = self::TT_PROHIBITED;
 169                          break;
 170  
 171                      case '~':
 172                          $this->type = self::TT_FUZZY_PROX_MARK;
 173                          break;
 174  
 175                      case '^':
 176                          $this->type = self::TT_BOOSTING_MARK;
 177                          break;
 178  
 179                      case '[':
 180                          $this->type = self::TT_RANGE_INCL_START;
 181                          break;
 182  
 183                      case ']':
 184                          $this->type = self::TT_RANGE_INCL_END;
 185                          break;
 186  
 187                      case '{':
 188                          $this->type = self::TT_RANGE_EXCL_START;
 189                          break;
 190  
 191                      case '}':
 192                          $this->type = self::TT_RANGE_EXCL_END;
 193                          break;
 194  
 195                      case '(':
 196                          $this->type = self::TT_SUBQUERY_START;
 197                          break;
 198  
 199                      case ')':
 200                          $this->type = self::TT_SUBQUERY_END;
 201                          break;
 202  
 203                      case '!':
 204                          $this->type = self::TT_NOT_LEXEME;
 205                          break;
 206  
 207                      case '&&':
 208                          $this->type = self::TT_AND_LEXEME;
 209                          break;
 210  
 211                      case '||':
 212                          $this->type = self::TT_OR_LEXEME;
 213                          break;
 214  
 215                      default:
 216                          throw new Zend_Search_Lucene_Exception('Unrecognized query syntax lexeme: \'' . $tokenText . '\'');
 217                  }
 218                  break;
 219  
 220              case self::TC_NUMBER:
 221                  $this->type = self::TT_NUMBER;
 222  
 223              default:
 224                  throw new Zend_Search_Lucene_Exception('Unrecognized lexeme type: \'' . $tokenCategory . '\'');
 225          }
 226      }
 227  }
 228