On Apr 2, 2004, at 10:03 AM, [EMAIL PROTECTED] wrote:
if (time != null) doc.add(Field.Keyword("date", new Date(time.getTime())));
query: date:[20030101 TO 20030202] date:20030101
The first query does throw a ParserException, the second doesn't return any
hits.
QueryParser uses DateFormat.SHORT to parse dates using the default locale. 20030101 does not parse using that format string.
If you are truly only representing dates (not times) then use, what Otis meant, YYYYMMDD *Strings* to represent the date. If you need times as well, first be aware of what RangeQuery does (expand all terms in that range!) so that you aren't shocked with performance or a too many clauses exception.
When you use Field.Keyword(String, Date), the date is converted into a lexicographically ordered value (the ugly thing you saw in Luke). This is why, if you only care about dates, YYYYMMDD is recommended (it must sort properly).
Hmm...there must be something simple I misunderstood :) BTW what about custom
date format in QueryParser (...and are the last two digits actually the day
or month)?
Custom date format in QueryParser is quite do-able. Subclass QueryParser and override getRangeQuery and do what you like there.
I do something similar in Lucene in Action. I'd like users to search on ranges like lastmodified:[1/1/04 TO 12/31/04], but internally I represent dates as YYYYMMDD. To make the conversion, I had to have a custom QueryParser....
/**
* From Lucene in Action (Manning Publications)
*/
public class SmartDayQueryParser extends QueryParser { public static final DateFormat formatter =
new SimpleDateFormat("yyyyMMdd"); public SmartDayQueryParser(String field, Analyzer analyzer) {
super(field, analyzer);
} protected Query getRangeQuery(String field, Analyzer analyzer,
String part1, String part2,
boolean inclusive)
throws ParseException {
try {
DateFormat df =
DateFormat.getDateInstance(DateFormat.SHORT,
getLocale());
df.setLenient(true);
Date d1 = df.parse(part1);
Date d2 = df.parse(part2);
part1 = formatter.format(d1);
part2 = formatter.format(d2);
} catch (Exception ignored) {
}return new RangeQuery(new Term(field, part1), new Term(field, part2), inclusive); } }
Erik
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
