2002-05-08 22:18:00 -04:00
< html >
< head >
2004-05-17 09:04:42 -04:00
< meta name = "Author" content = "Kelvin Tan" >
< title > Lucene Query Constructor Demo and Introduction< / title >
2002-05-08 22:18:00 -04:00
< script type = "text/javascript" src = "luceneQueryConstructor.js" > < / script >
2004-05-17 09:29:24 -04:00
< script type = "text/javascript" src = "../queryValidator/luceneQueryValidator.js" > < / script >
2004-05-17 09:04:42 -04:00
< script >
2004-05-17 09:29:24 -04:00
submitForm = false // necessary for luceneQueryConstructor not to submit the form upon query construction
function doSubmitForm(frm)
2004-05-17 09:04:42 -04:00
{
if(frm["noField-phrase-input"].value.length > 0)
frm["noField-phrase"].value = quote(frm["noField-phrase-input"].value)
else if(frm["noField-phrase"].value.length > 0)
frm["noField-phrase"].value = ''
2004-05-17 09:29:24 -04:00
doMakeQuery(frm.query);
2004-05-17 09:04:42 -04:00
}
< / script >
2002-05-08 22:18:00 -04:00
< / head >
2004-05-17 09:04:42 -04:00
2002-05-08 22:18:00 -04:00
< body >
2004-05-17 09:04:42 -04:00
< h2 > Lucene Javascript Query Constructor< / h2 >
< p > luceneQueryConstructor.js is a Javascript framework for constructing queries using the "advanced" search features of lucene,
namely field-searching, boolean searching, phrase searching, group searching (via parentheses) and various combinations of the aforementioned.
< p > It also provides a convenient way to mimic a Google-like search, where all terms are ANDed, as opposed to Lucene's default OR modifier.
< p >
This HTML form provides examples on the usage of luceneQueryConstructor.js.
An interface similar to < a href = "http://www.google.com/advanced_search" > Google's Advanced Search< / a > form is shown here.
2002-05-08 22:18:00 -04:00
< form >
2004-05-17 09:04:42 -04:00
< table width = "100%" border = "0" cellspacing = "1" cellpadding = "5" >
< tr >
< th > < / th >
< td width = "25%" > < / td >
< tr >
< th >
< input name = "noField-andModifier" value = "+|0" type = "hidden" > < b > Find results< / b >
< / th >
< td class = "bodytext" > With < b > all< / b > of the words< / td >
< td class = "bodytext" >
< input type = "text" name = "noField-and" size = "25" >
< select name = "resultsPerPage" >
< option value = "10" > 10 results< option value = "20" > 20 results
< option value = "50" selected > 50 results< / select >
< / td >
< / tr >
< tr >
< th >
< input name = "noField-phraseModifier" value = "+|+" type = "hidden" >
< / th >
< td class = "bodytext" > With the < b > exact phrase< / b > < / td >
< td class = "bodytext" >
< input type = "text" name = "noField-phrase-input" size = "25" >
< input type = "hidden" name = "noField-phrase" >
< / td >
< / tr >
< tr >
< th >
< input name = "noField-orModifier" value = " |+" type = "hidden" >
< / th >
< td class = "bodytext" > With < b > at least< / b > one of the words< / td >
< td class = "bodytext" >
< input type = "text" name = "noField-or" size = "25" >
< / td >
< / tr >
< tr >
< th >
< input name = "noField-notModifier" value = "-|0" type = "hidden" >
< / th >
< td class = "bodytext" > < b > Without< / b > the words< / td >
< td class = "bodytext" >
< input type = "text" name = "noField-not" size = "25" >
< / td >
< / tr >
< tr >
< th >
< b > File Format< / b >
< / th >
< td class = "bodytext" >
< select name = "fileNameModifier" > < option value = "And" selected > Only< / option > < option value = "Not" > Don't< / option > < / select >
return results of the file format< / td >
< td class = "bodytext" >
< select name = "fileName" > < option value = "" selected > any format< option value = "pdf" > Adobe Acrobat PDF (.pdf)
< option value = "doc" > Microsoft Word (.doc)< option value = "xls" > Microsoft Excel (.xls)< option value = "ppt" > Microsoft Powerpoint (.ppt)< / select >
< / td >
< / tr >
< tr >
< th >
< b > Date< / b >
< / th >
< td class = "bodytext" >
Return results updated in the < / td >
< td class = "bodytext" >
< select name = "date" > < option value = "" selected > anytime< option value = "3" > past 3 months
< option value = "6" > past 6 months< option value = "12" > past year< / select >
< input type = "hidden" name = "fromDate" >
< input type = "hidden" name = "toDate" >
< input type = "hidden" name = "dateRangeField" value = "lastModifiedDate" >
< / td >
< / tr >
2004-05-17 09:29:24 -04:00
< tr >
2002-05-08 22:18:00 -04:00
< input type = "hidden" name = "query" >
2004-05-17 09:29:24 -04:00
< tr > < td > < / tr >
< tr > < th > < p > Current Query:< / th > < td > < pre id = "curQuery" > < / pre > < pre id = "curQueryValid" > < / pre > < / td > < td >
< input type = "button" name = "Update" value = "Update Query" onClick = "doSubmitForm(this.form); curQuery.innerText = this.form.query.value" / >
< input type = "button" name = "Validate" value = "Validate" onClick = "doCheckLuceneQuery(this.form.query); curQueryValid.innerText = 'Query is valid'" / >
< / td >
2004-05-17 09:04:42 -04:00
< / table >
2002-05-08 22:18:00 -04:00
< / form >
2004-05-17 09:04:42 -04:00
< p > luceneQueryConstructor works by assuming a certain naming convention of form fields to obtain the necessary information to construct the query.< br >
< b > NB:< / b > Unless otherwise specified, all uses of the word < code > field< / code > should be assumed to mean form input fields and not Lucene document fields.
< p > The input form field is expected to be the < b > same name as the Lucene Field< / b > . For example, if you have a Document with < i > fileName< / i > as a Field, and
you'd like to provide field-searching on this field, then introduce a form field like so:
< center >
< p > < code > < input type="text" name="fileName"> < / code >
< / p >
< / center >
You are also expected to provide another field known as
this field's < b > modifier< / b > . This modifier field tells luceneQueryConstructor how to convert the field and value into a Lucene query. The naming convention
of the modifier is < b > < name of input field/Lucene field> < modifier suffix as declared in luceneQueryConstructor.js> < / b > . So, for the < i > fileName< / i > field
we introduced above, it's modifier field would be:
< center >
< p > < code > < input type="hidden" name="fileNameModifier" value="+|+"> < / code >
< / p >
< / center >
< p > The value of the modifier field is in the form < b > < term modifier> |< group modifier> < / b > . Let me explain.
< p > Looking at the form above, we see fields that provide
< ol >
< li > AND search
< li > OR search
< li > NOT search
< li > and others which are unrelated to this discussion
< / ol >
Given a value of < b > < i > foo bar< / i > < / b > , the AND search field must be converted to < b > < i > +foo +bar< / i > < / b > (luceneQueryConstructor only supports
using +/-, not AND/OR/NOT), the NOT search to < b > < i > -foo -bar< / i > < / b > and the OR search not at all.
< p > However, also consider the relationship < b > between< / b > these groups of fields. Assuming Google's Advanced Search interface,
we're effectively saying that we want all of the terms in the AND search field < b > AND< / b > at least one of the
terms in the OR search field < b > AND< / b > none of the terms in the NOT search.
< p > So, if the AND, OR and NOT search fields all have the values of < b > < i > foo bar< / i > < / b > , then an appropriate search query
which fulfills the requirements would be
< center >
< p > < code > +foo +bar +(foo bar) -foo -bar< / code >
< / p >
< / center >
Well, to be more correct, it should be
< center >
< p > < code > +(+foo +bar) +(foo bar) -foo -bar< / code >
< / p >
< / center >
Hmmmm...if you're sharp, you would have noticed that the NOT terms aren't grouped.
You'll find that if you group them with an AND modifier, no results will be returned at all (though it's a valid query),
because the query constructed wouldn't make any sense at all. Lucene also implicitly ANDs NOT terms, it seems. In any case,
both queries as presented are correct, though I prefer the first one because it is less verbose.
< p > The following matrix provides modifiers and their effects on queries:< br > < br >
< table border = "1" align = "center" >
< tr >
< th > Boolean modifier
< th > Form value
< th > As term modifier
< th > As group modifier
< / tr >
< tr align = "center" >
< td > AND
< td > +
< td > +term1 + term2 ...
< td > +(...)
< / tr >
< tr align = "center" >
< td > OR
< td > (single space)
< td > term1 term2 ...
< td > (...)
< / tr >
< tr align = "center" >
< td > NOT
< td > -
< td > -term1 -term2 ...
< td > -(...)
< / tr >
< tr align = "center" >
< td > no modifier
< td > 0
< td > term1 term2 ...
< td > no grouping
< / tr >
< / table >
< p > With this knowledge, we know that the value of the AND field modifier needs to be < b > < i > +|0< / i > < / b > for the first query and
< b > < i > +|+< / i > < / b > for the second query, the values of the NOT field modifier and the
OR field modifier are < b > < i > -|0< / i > < / b > and < b > < i > |+< / i > < / b > (it's an empty space before the |) in both queries respectively.
< p >
Well, that's all I have to say for now. There are more topics to be covered, such as construction of phrase searches, non-field searches,
multiple list box selections, radio buttons etc, but right now I'm not even sure if anyone will read this much! :-) Anyway, there's always the code.
2002-05-08 22:18:00 -04:00
< / body >
< / html >
