querylexer.php

很棒的在线教学系统

<?php
/**
 * Zend Framework
 *
 * LICENSE
 *
 * This source file is subject to the new BSD license that is bundled
 * with this package in the file LICENSE.txt.
 * It is also available through the world-wide-web at this URL:
 * http://framework.zend.com/license/new-bsd
 * If you did not receive a copy of the license and are unable to
 * obtain it through the world-wide-web, please send an email
 * to [e-mail] so we can send you a copy immediately.
 *
 * @category   Zend
 * @package    Zend_Search_Lucene
 * @subpackage Search
 * @copyright  Copyright (c) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
 * @license    http://framework.zend.com/license/new-bsd     New BSD License
 */


/** Zend_Search_Lucene_FSM */
require_once 'Zend/Search/Lucene/FSM.php';

/** Zend_Search_Lucene_Search_QueryParser */
require_once 'Zend/Search/Lucene/Search/QueryToken.php';

/** Zend_Search_Lucene_Exception */
require_once 'Zend/Search/Lucene/Exception.php';

/** Zend_Search_Lucene_Search_QueryParserException */
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';


/**
 * @category   Zend
 * @package    Zend_Search_Lucene
 * @subpackage Search
 * @copyright  Copyright (c) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
 * @license    http://framework.zend.com/license/new-bsd     New BSD License
 */
class Zend_Search_Lucene_Search_QueryLexer extends Zend_Search_Lucene_FSM
{
    /** State Machine states */
    const ST_WHITE_SPACE     = 0;
    const ST_SYNT_LEXEME     = 1;
    const ST_LEXEME          = 2;
    const ST_QUOTED_LEXEME   = 3;
    const ST_ESCAPED_CHAR    = 4;
    const ST_ESCAPED_QCHAR   = 5;
    const ST_LEXEME_MODIFIER = 6;
    const ST_NUMBER          = 7;
    const ST_MANTISSA        = 8;
    const ST_ERROR           = 9;

    /** Input symbols */
    const IN_WHITE_SPACE     = 0;
    const IN_SYNT_CHAR       = 1;
    const IN_LEXEME_MODIFIER = 2;
    const IN_ESCAPE_CHAR     = 3;
    const IN_QUOTE           = 4;
    const IN_DECIMAL_POINT   = 5;
    const IN_ASCII_DIGIT     = 6;
    const IN_CHAR            = 7;
    const IN_MUTABLE_CHAR    = 8;

    const QUERY_WHITE_SPACE_CHARS      = " \n\r\t";
    const QUERY_SYNT_CHARS             = ':()[]{}!|&';
    const QUERY_MUTABLE_CHARS          = '+-';
    const QUERY_DOUBLECHARLEXEME_CHARS = '|&';
    const QUERY_LEXEMEMODIFIER_CHARS   = '~^';
    const QUERY_ASCIIDIGITS_CHARS      = '0123456789';

    /**
     * List of recognized lexemes
     *
     * @var array
     */
    private $_lexemes;

    /**
     * Query string (array of single- or non single-byte characters)
     *
     * @var array
     */
    private $_queryString;

    /**
     * Current position within a query string
     * Used to create appropriate error messages
     *
     * @var integer
     */
    private $_queryStringPosition;

    /**
     * Recognized part of current lexeme
     *
     * @var string
     */
    private $_currentLexeme;

    public function __construct()
    {
        parent::__construct( array(self::ST_WHITE_SPACE,
                                   self::ST_SYNT_LEXEME,
                                   self::ST_LEXEME,
                                   self::ST_QUOTED_LEXEME,
                                   self::ST_ESCAPED_CHAR,
                                   self::ST_ESCAPED_QCHAR,
                                   self::ST_LEXEME_MODIFIER,
                                   self::ST_NUMBER,
                                   self::ST_MANTISSA,
                                   self::ST_ERROR),
                             array(self::IN_WHITE_SPACE,
                                   self::IN_SYNT_CHAR,
                                   self::IN_MUTABLE_CHAR,
                                   self::IN_LEXEME_MODIFIER,
                                   self::IN_ESCAPE_CHAR,
                                   self::IN_QUOTE,
                                   self::IN_DECIMAL_POINT,
                                   self::IN_ASCII_DIGIT,
                                   self::IN_CHAR));


        $lexemeModifierErrorAction    = new Zend_Search_Lucene_FSMAction($this, 'lexModifierErrException');
        $quoteWithinLexemeErrorAction = new Zend_Search_Lucene_FSMAction($this, 'quoteWithinLexemeErrException');
        $wrongNumberErrorAction       = new Zend_Search_Lucene_FSMAction($this, 'wrongNumberErrException');



        $this->addRules(array( array(self::ST_WHITE_SPACE,   self::IN_WHITE_SPACE,     self::ST_WHITE_SPACE),
                               array(self::ST_WHITE_SPACE,   self::IN_SYNT_CHAR,       self::ST_SYNT_LEXEME),
                               array(self::ST_WHITE_SPACE,   self::IN_MUTABLE_CHAR,    self::ST_SYNT_LEXEME),
                               array(self::ST_WHITE_SPACE,   self::IN_LEXEME_MODIFIER, self::ST_LEXEME_MODIFIER),
                               array(self::ST_WHITE_SPACE,   self::IN_ESCAPE_CHAR,     self::ST_ESCAPED_CHAR),
                               array(self::ST_WHITE_SPACE,   self::IN_QUOTE,           self::ST_QUOTED_LEXEME),
                               array(self::ST_WHITE_SPACE,   self::IN_DECIMAL_POINT,   self::ST_LEXEME),
                               array(self::ST_WHITE_SPACE,   self::IN_ASCII_DIGIT,     self::ST_LEXEME),
                               array(self::ST_WHITE_SPACE,   self::IN_CHAR,            self::ST_LEXEME)
                             ));
        $this->addRules(array( array(self::ST_SYNT_LEXEME,   self::IN_WHITE_SPACE,     self::ST_WHITE_SPACE),
                               array(self::ST_SYNT_LEXEME,   self::IN_SYNT_CHAR,       self::ST_SYNT_LEXEME),
                               array(self::ST_SYNT_LEXEME,   self::IN_MUTABLE_CHAR,    self::ST_SYNT_LEXEME),
                               array(self::ST_SYNT_LEXEME,   self::IN_LEXEME_MODIFIER, self::ST_LEXEME_MODIFIER),
                               array(self::ST_SYNT_LEXEME,   self::IN_ESCAPE_CHAR,     self::ST_ESCAPED_CHAR),
                               array(self::ST_SYNT_LEXEME,   self::IN_QUOTE,           self::ST_QUOTED_LEXEME),
                               array(self::ST_SYNT_LEXEME,   self::IN_DECIMAL_POINT,   self::ST_LEXEME),
                               array(self::ST_SYNT_LEXEME,   self::IN_ASCII_DIGIT,     self::ST_LEXEME),
                               array(self::ST_SYNT_LEXEME,   self::IN_CHAR,            self::ST_LEXEME)
                             ));
        $this->addRules(array( array(self::ST_LEXEME,        self::IN_WHITE_SPACE,     self::ST_WHITE_SPACE),
                               array(self::ST_LEXEME,        self::IN_SYNT_CHAR,       self::ST_SYNT_LEXEME),
                               array(self::ST_LEXEME,        self::IN_MUTABLE_CHAR,    self::ST_LEXEME),
                               array(self::ST_LEXEME,        self::IN_LEXEME_MODIFIER, self::ST_LEXEME_MODIFIER),
                               array(self::ST_LEXEME,        self::IN_ESCAPE_CHAR,     self::ST_ESCAPED_CHAR),

                               // IN_QUOTE     not allowed
                               array(self::ST_LEXEME,        self::IN_QUOTE,           self::ST_ERROR, $quoteWithinLexemeErrorAction),

                               array(self::ST_LEXEME,        self::IN_DECIMAL_POINT,   self::ST_LEXEME),
                               array(self::ST_LEXEME,        self::IN_ASCII_DIGIT,     self::ST_LEXEME),
                               array(self::ST_LEXEME,        self::IN_CHAR,            self::ST_LEXEME)
                             ));
        $this->addRules(array( array(self::ST_QUOTED_LEXEME, self::IN_WHITE_SPACE,     self::ST_QUOTED_LEXEME),
                               array(self::ST_QUOTED_LEXEME, self::IN_SYNT_CHAR,       self::ST_QUOTED_LEXEME),
                               array(self::ST_QUOTED_LEXEME, self::IN_MUTABLE_CHAR,    self::ST_QUOTED_LEXEME),
                               array(self::ST_QUOTED_LEXEME, self::IN_LEXEME_MODIFIER, self::ST_QUOTED_LEXEME),
                               array(self::ST_QUOTED_LEXEME, self::IN_ESCAPE_CHAR,     self::ST_ESCAPED_QCHAR),
                               array(self::ST_QUOTED_LEXEME, self::IN_QUOTE,           self::ST_WHITE_SPACE),
                               array(self::ST_QUOTED_LEXEME, self::IN_DECIMAL_POINT,   self::ST_QUOTED_LEXEME),
                               array(self::ST_QUOTED_LEXEME, self::IN_ASCII_DIGIT,     self::ST_QUOTED_LEXEME),
                               array(self::ST_QUOTED_LEXEME, self::IN_CHAR,            self::ST_QUOTED_LEXEME)
                             ));
        $this->addRules(array( array(self::ST_ESCAPED_CHAR,  self::IN_WHITE_SPACE,     self::ST_LEXEME),
                               array(self::ST_ESCAPED_CHAR,  self::IN_SYNT_CHAR,       self::ST_LEXEME),
                               array(self::ST_ESCAPED_CHAR,  self::IN_MUTABLE_CHAR,    self::ST_LEXEME),
                               array(self::ST_ESCAPED_CHAR,  self::IN_LEXEME_MODIFIER, self::ST_LEXEME),
                               array(self::ST_ESCAPED_CHAR,  self::IN_ESCAPE_CHAR,     self::ST_LEXEME),
                               array(self::ST_ESCAPED_CHAR,  self::IN_QUOTE,           self::ST_LEXEME),
                               array(self::ST_ESCAPED_CHAR,  self::IN_DECIMAL_POINT,   self::ST_LEXEME),
                               array(self::ST_ESCAPED_CHAR,  self::IN_ASCII_DIGIT,     self::ST_LEXEME),
                               array(self::ST_ESCAPED_CHAR,  self::IN_CHAR,            self::ST_LEXEME)
                             ));
        $this->addRules(array( array(self::ST_ESCAPED_QCHAR, self::IN_WHITE_SPACE,     self::ST_QUOTED_LEXEME),
                               array(self::ST_ESCAPED_QCHAR, self::IN_SYNT_CHAR,       self::ST_QUOTED_LEXEME),
                               array(self::ST_ESCAPED_QCHAR, self::IN_MUTABLE_CHAR,    self::ST_QUOTED_LEXEME),
                               array(self::ST_ESCAPED_QCHAR, self::IN_LEXEME_MODIFIER, self::ST_QUOTED_LEXEME),
                               array(self::ST_ESCAPED_QCHAR, self::IN_ESCAPE_CHAR,     self::ST_QUOTED_LEXEME),
                               array(self::ST_ESCAPED_QCHAR, self::IN_QUOTE,           self::ST_QUOTED_LEXEME),
                               array(self::ST_ESCAPED_QCHAR, self::IN_DECIMAL_POINT,   self::ST_QUOTED_LEXEME),
                               array(self::ST_ESCAPED_QCHAR, self::IN_ASCII_DIGIT,     self::ST_QUOTED_LEXEME),
                               array(self::ST_ESCAPED_QCHAR, self::IN_CHAR,            self::ST_QUOTED_LEXEME)
                             ));
        $this->addRules(array( array(self::ST_LEXEME_MODIFIER, self::IN_WHITE_SPACE,     self::ST_WHITE_SPACE),
                               array(self::ST_LEXEME_MODIFIER, self::IN_SYNT_CHAR,       self::ST_SYNT_LEXEME),
                               array(self::ST_LEXEME_MODIFIER, self::IN_MUTABLE_CHAR,    self::ST_SYNT_LEXEME),
                               array(self::ST_LEXEME_MODIFIER, self::IN_LEXEME_MODIFIER, self::ST_LEXEME_MODIFIER),

                               // IN_ESCAPE_CHAR       not allowed
                               array(self::ST_LEXEME_MODIFIER, self::IN_ESCAPE_CHAR,     self::ST_ERROR, $lexemeModifierErrorAction),

                               // IN_QUOTE             not allowed
                               array(self::ST_LEXEME_MODIFIER, self::IN_QUOTE,           self::ST_ERROR, $lexemeModifierErrorAction),


                               array(self::ST_LEXEME_MODIFIER, self::IN_DECIMAL_POINT,   self::ST_MANTISSA),
                               array(self::ST_LEXEME_MODIFIER, self::IN_ASCII_DIGIT,     self::ST_NUMBER),

                               // IN_CHAR              not allowed
                               array(self::ST_LEXEME_MODIFIER, self::IN_CHAR,            self::ST_ERROR, $lexemeModifierErrorAction),
                             ));
        $this->addRules(array( array(self::ST_NUMBER, self::IN_WHITE_SPACE,     self::ST_WHITE_SPACE),
                               array(self::ST_NUMBER, self::IN_SYNT_CHAR,       self::ST_SYNT_LEXEME),
                               array(self::ST_NUMBER, self::IN_MUTABLE_CHAR,    self::ST_SYNT_LEXEME),
                               array(self::ST_NUMBER, self::IN_LEXEME_MODIFIER, self::ST_LEXEME_MODIFIER),

                               // IN_ESCAPE_CHAR       not allowed
                               array(self::ST_NUMBER, self::IN_ESCAPE_CHAR,     self::ST_ERROR, $wrongNumberErrorAction),

                               // IN_QUOTE             not allowed
                               array(self::ST_NUMBER, self::IN_QUOTE,           self::ST_ERROR, $wrongNumberErrorAction),

                               array(self::ST_NUMBER, self::IN_DECIMAL_POINT,   self::ST_MANTISSA),
                               array(self::ST_NUMBER, self::IN_ASCII_DIGIT,     self::ST_NUMBER),

                               // IN_CHAR              not allowed
                               array(self::ST_NUMBER, self::IN_CHAR,            self::ST_ERROR, $wrongNumberErrorAction),
                             ));
        $this->addRules(array( array(self::ST_MANTISSA, self::IN_WHITE_SPACE,     self::ST_WHITE_SPACE),
                               array(self::ST_MANTISSA, self::IN_SYNT_CHAR,       self::ST_SYNT_LEXEME),
                               array(self::ST_MANTISSA, self::IN_MUTABLE_CHAR,    self::ST_SYNT_LEXEME),
                               array(self::ST_MANTISSA, self::IN_LEXEME_MODIFIER, self::ST_LEXEME_MODIFIER),

                               // IN_ESCAPE_CHAR       not allowed
                               array(self::ST_MANTISSA, self::IN_ESCAPE_CHAR,     self::ST_ERROR, $wrongNumberErrorAction),

                               // IN_QUOTE             not allowed
                               array(self::ST_MANTISSA, self::IN_QUOTE,           self::ST_ERROR, $wrongNumberErrorAction),

                               // IN_DECIMAL_POINT     not allowed
                               array(self::ST_MANTISSA, self::IN_DECIMAL_POINT,   self::ST_ERROR, $wrongNumberErrorAction),

                               array(self::ST_MANTISSA, self::IN_ASCII_DIGIT,     self::ST_MANTISSA),

                               // IN_CHAR              not allowed
                               array(self::ST_MANTISSA, self::IN_CHAR,            self::ST_ERROR, $wrongNumberErrorAction),
                             ));


        /** Actions */
        $syntaxLexemeAction    = new Zend_Search_Lucene_FSMAction($this, 'addQuerySyntaxLexeme');
        $lexemeModifierAction  = new Zend_Search_Lucene_FSMAction($this, 'addLexemeModifier');
        $addLexemeAction       = new Zend_Search_Lucene_FSMAction($this, 'addLexeme');