class.ilLuceneQueryParser.php

Go to the documentation of this file.

<?php
 
declare(strict_types=1);
class ilLuceneQueryParser
{
    protected string $query_string;
    protected string $parsed_query = '';
 
 
    public function __construct($a_query_string)
    {
        $this->query_string = $a_query_string;
    }
 
    public function parse(): void
    {
        $this->parsed_query = (string) preg_replace_callback(
            '/(owner:)\s?([A-Za-z0-9_\.\+\*\@!\$\%\~\-]+)/',
            array($this,'replaceOwnerCallback'),
            $this->query_string
        );
    }
 
    public function parseAutoWildcard(): void
    {
        $this->parsed_query = trim($this->query_string);
        if (stristr($this->parsed_query, '*')) {
            return;
        }
        if (substr($this->parsed_query, -1) !== '"') {
            $this->parsed_query .= '*';
        }
    }
 
    public function getQuery(): string
    {
        return $this->parsed_query;
    }
 
    protected function replaceOwnerCallback(array $matches): string
    {
        if (isset($matches[2])) {
            if ($usr_id = ilObjUser::_loginExists($matches[2])) {
                return $matches[1] . $usr_id;
            }
        }
        return $matches[0];
    }
 
 
    public static function validateQuery($a_query): bool
    {
        #ilLuceneQueryParser::checkAllowedCharacters($a_query);
        #ilLuceneQueryParser::checkAsterisk($a_query);
        #ilLuceneQueryParser::checkAmpersands($a_query);
        ilLuceneQueryParser::checkCaret($a_query);
        ilLuceneQueryParser::checkSquiggle($a_query);
        #ilLuceneQueryParser::checkExclamationMark($a_query);
        #ilLuceneQueryParser::checkQuestionMark($a_query);
        ilLuceneQueryParser::checkParenthesis($a_query);
        #ilLuceneQueryParser::checkPlusMinus($a_query);
        #ilLuceneQueryParser::checkANDORNOT($a_query);
        ilLuceneQueryParser::checkQuotes($a_query);
        #ilLuceneQueryParser::checkColon($a_query);
        return true;
    }
 
    protected static function checkAllowedCharacters(string $query): bool
    {
        if (preg_match('/[^\pL0-9_+\-:.()\"*?&§€|!{}\[\]\^~\\@#\/$%\'= ]/u', $query) != 0) {
            throw new ilLuceneQueryParserException('lucene_err_allowed_characters');
        }
        return true;
    }
 
    protected static function checkAsterisk(string $query): bool
    {
        if (preg_match('/^[\*]*$|[\s]\*|^\*[^\s]/', $query) != 0) {
            throw new ilLuceneQueryParserException('lucene_err_asterisk');
        }
        return true;
    }
 
    protected static function checkAmpersands(string $query): bool
    {
        if (preg_match('/[&]{2}/', $query) > 0) {
            if (preg_match('/^([\pL0-9_+\-:.()\"*?&|!{}\[\]\^~\\@#\/$%\'=]+( && )?[\pL0-9_+\-:.()\"*?|!{}\[\]\^~\\@#\/$%\'=]+[ ]*)+$/u', $query) == 0) {
                throw new ilLuceneQueryParserException('lucene_err_ampersand');
            }
        }
        return true;
    }
 
    protected static function checkCaret(string $query): bool
    {
        if (preg_match('/[^\\\]\^([^\s]*[^0-9.]+)|[^\\\]\^$/', $query) != 0) {
            throw new ilLuceneQueryParserException('lucene_err_caret');
        }
        return true;
    }
 
    protected static function checkSquiggle(string $query): bool
    {
        if (preg_match('/[^\\\]*~[^\s]*[^0-9\s]+/', $query, $matches) != 0) {
            throw new ilLuceneQueryParserException('lucene_err_squiggle');
        }
        return true;
    }
 
    protected static function checkExclamationMark(string $query): bool
    {
        if (preg_match('/^[^!]*$|^([\pL0-9_+\-:.()\"*?&|!{}\[\]\^~\\@#\/$%\'=]+( ! )?[\pL0-9_+\-:.()\"*?&|!{}\[\]\^~\\@#\/$%\'=]+[ ]*)+$/u', $query, $matches) == 0) {
            throw new ilLuceneQueryParserException('lucene_err_exclamation_mark');
        }
        return true;
    }
 
    protected static function checkQuestionMark(string $query): bool
    {
        if (preg_match('/^(\?)|([^\pL0-9_+\-:.()\"*?&|!{}\[\]\^~\\@#\/$%\'=]\?+)/u', $query, $matches) != 0) {
            throw new ilLuceneQueryParserException('lucene_err_question_mark');
        }
        return true;
    }
 
    protected static function checkParenthesis(string $a_query): bool
    {
        $hasLft = false;
        $hasRgt = false;
 
        $matchLft = 0;
        $matchRgt = 0;
 
        $tmp = array();
 
        if (($matchLft = preg_match_all('/[(]/', $a_query, $tmp)) > 0) {
            $hasLft = true;
        }
        if (($matchRgt = preg_match_all('/[)]/', $a_query, $tmp)) > 0) {
            $hasRgt = true;
        }
 
        if (!$hasLft || !$hasRgt) {
            return true;
        }
 
 
        if (($hasLft && !$hasRgt) || ($hasRgt && !$hasLft)) {
            throw new ilLuceneQueryParserException('lucene_err_parenthesis_not_closed');
        }
 
        if ($matchLft !== $matchRgt) {
            throw new ilLuceneQueryParserException('lucene_err_parenthesis_not_closed');
        }
 
        if (preg_match('/\‍(\s*\‍)/', $a_query) > 0) {
            throw new ilLuceneQueryParserException('lucene_err_parenthesis_empty');
        }
        return true;
    }
 
    protected static function checkPlusMinus(string $a_query): bool
    {
        if (preg_match('/^[^\n+\-]*$|^([+-]?\s*[\pL0-9_:.()\"*?&|!{}\[\]\^~\\@#\/$%\'=]+[ ]?)+$/u', $a_query) == 0) {
            throw new ilLuceneQueryParserException('lucene_err_plus_minus');
        }
        return true;
    }
 
    protected static function checkANDORNOT(string $a_query): bool
    {
        if (preg_match('/^([\pL0-9_+\-:.()\"*?&|!{}\[\]\^~\\@\/#$%\'=]+\s*((AND )|(OR )|(AND NOT )|(NOT ))?[\pL0-9_+\-:.()\"*?&|!{}\[\]\^~\\@\/#$%\'=]+[ ]*)+$/u', $a_query) == 0) {
            throw new ilLuceneQueryParserException('lucene_err_and_or_not');
        }
        return true;
    }
 
    protected static function checkQuotes(string $a_query): bool
    {
        $matches = preg_match_all('/"/', $a_query, $tmp);
 
        if ($matches == 0) {
            return true;
        }
 
        if (($matches % 2) > 0) {
            throw new ilLuceneQueryParserException('lucene_err_quotes');
        }
 
        if (preg_match('/"\s*"/', $a_query) > 0) {
            throw new ilLuceneQueryParserException('lucene_err_quotes_not_empty');
        }
        return true;
    }
 
 
    protected static function checkColon(string $a_query): bool
    {
        if (preg_match('/[^\\\\s]:[\s]|[^\\\\s]:$|[\s][^\\]?:|^[^\\\\s]?:/', $a_query) != 0) {
            throw new ilLuceneQueryParserException('lucene_err_colon');
        }
        return true;
    }
}