Zend_Search_Lucene range query error

I have set up each document with a date field. (keyword)

Values stored in it are in this format; 20100511

Each time I try to perform a ranged query, I get the following error:

date:[10000000 TO 20000000]

At least one range query boundary term must be non-empty te开发者_JAVA技巧rm

Anyone got a clue?

Update

I have gotten this to work programmatically. Does that mean the parser is buggy?

$from  = new Zend_Search_Lucene_Index_Term('10000000', 'dateOfBirthMod');
$to    = new Zend_Search_Lucene_Index_Term('20000000', 'dateOfBirthMod');
$query = new Zend_Search_Lucene_Search_Query_Range($from, $to, true);

---

Actually, this is more of a questionable default, not a bug. You can change the analyzer to allow numbers. In fact, you can even write a custom analyzer. See http://framework.zend.com/manual/en/zend.search.lucene.extending.html

The setting for allowing numbers to be tokenized is

Zend_Search_Lucene_Analysis_Analyzer::setDefault(new Zend_Search_Lucene_Analysis_Analyzer_Common_TextNum_CaseInsensitive());

in Zf 1.x and in Zf 2.x

Zend\Search\Lucene\Analysis\Analyzer\Analyzer::setDefault(new Zend\Search\Lucene\Analysis\Analyzer\Common\TextNum\CaseInsensitive());

---

Apparently it is a bug in the query parser (quite old btw). I would suggest that you either add a comment to that issue or open a new one to confirm that it is still happening in version x.x of the ZF.

---

I've made a workaround for this bug which stems from the method called tokenize() which does not return any value and which may be found in the Zend/Search/Lucene/Analysis/Analyzer.php

You can try to replace the code with the following one if you use the latest ZF release (1.10.7).

public function tokenize($data, $encoding = '')
{
    $this->setInput($data, $encoding);

    $tokenList = array();
    /*
    while (($nextToken = $this->nextToken()) !== null) {
        $tokenList[] = $this->_input;
    }
    */
        $tokenList[] = new Zend_Search_Lucene_Analysis_Token( $this->_input, 1, 1 );

    return $tokenList;
}

I don't know whether it works in older releases or not.

---

First, you have to change default Analyzer to TextNum to allow number before finding

ZF2:

\ZendSearch\Lucene\Analysis\Analyzer\Analyzer::setDefault(new \ZendSearch\Lucene\Analysis\Analyzer\Common\TextNum\CaseInsensitive());

Second, your date field MUST be a keyword, not text.