Lucene - Sorting Date as NumericField - string

Question

While trying to sort datetime (long) numeric fields I always get a FormatException.
When converting a string to DateTime, parse the string to take the
date before putting each variable into the DateTime object.
Adding the numeric field:
doc.Add(new NumericField("creationDate", Field.Store.YES, true)
.SetLongValue(DateTime.UtcNow.Ticks);
Add sorting:
// boolean query
var sortField = new SortField("creationDate", SortField.LONG, true);
var inverseSort = new Sort(sortField);
var results = searcher.Search(query, null, 100, inverseSort); // exception thrown here
Inspecting the index, I can verify that 'creationDate' field is storing "long" values. What could be causing this exception?
EDIT:
Query
var query = new BooleanQuery();
foreach (var termQuery in incomingProps.Select(p => new TermQuery(new Term(kvp.Key, kvp.Value.ToLowerInvariant()))
{
query.Add(new BooleanClause(termQuery , Occur.Must));
}
return query;
Version: Lucene.Net 3.0.3
UPDATE:
This issue is occurring again, now with INT values.
I downloaded Lucene.Net source code and debugged the issue.
So it's somewhere in the FieldCache, when trying to parse the value "`\b\0\0\0" to Integer, which seems a bit odd.
I'm adding these values as numeric fields:
doc.Add(new NumericField(VersionNum, int.MaxValue, Field.Store.YES,
true).SetIntValue(VersionValue));
I get the exception when I'm supposed to get at least 1 hit back.
After inspecting the Index I see that the field's term is as following:
And the field text is:
EDIT:
I've hardcoded an int value and added a few segments:
doc.Add(new Field(VersionNum, NumericUtils.IntToPrefixCoded(1), Field.Store.YES, Field.Index.NOT_ANALYZED_NO_NORMS));
Which resulted on storing the version field as:
And still, when I try to sort I get the parsing error:
var sortVersion = new SortField(VersionNum, SortField.INT, true);
For every exception, Lucene is trying to parse " \b\0\0\0 ".
Looking at the prefixed coded stored as string, 1 would translate to " \b\0\0\0\1 " I'm guessing?
Is Lucene probably leaving some garbage behind in the FieldCache ?

Answer 1

Here's a unit test that tries to capture what you're asking. The test passes. Can you explain what the difference with your code is? (posting a full failing test would help us understand what you're doing :-) )
using System;
using System.Linq;
using System.Collections.Generic;
using Microsoft.VisualStudio.TestTools.UnitTesting;
using Lucene.Net.Search;
using Lucene.Net.Index;
using Lucene.Net.Analysis.Standard;
using Lucene.Net.QueryParsers;
using Lucene.Net.Documents;
using Lucene.Net.Store;
namespace SO_answers
{
[TestClass]
public class UnitTest1
{
[TestMethod]
public void TestShopping()
{
var item = new Dictionary<string, string>
{
{"field1", "value1" },
{"field2", "value2" },
{"field3", "value3" }
};
var writer = CreateIndex();
Add(writer, item);
writer.Flush(true, true, true);
var searcher = new IndexSearcher(writer.GetReader());
var result = Search(searcher, item);
Assert.AreEqual(1, result.Count);
writer.Dispose();
}
private List<string> Search(IndexSearcher searcher, Dictionary<string, string> values)
{
var query = new BooleanQuery();
foreach (var termQuery in values.Select(kvp => new TermQuery(new Term(kvp.Key, kvp.Value.ToLowerInvariant()))))
query.Add(new BooleanClause(termQuery, Occur.MUST));
return Search(searcher, query);
}
private List<string> Search(IndexSearcher searcher, Query query)
{
var sortField = new SortField("creationDate", SortField.LONG, true);
var inverseSort = new Sort(sortField);
var results = searcher.Search(query, null, 100, inverseSort); // exception thrown here
var result = new List<string>();
var matches = results.ScoreDocs;
foreach (var item in matches)
{
var id = item.Doc;
var doc = searcher.Doc(id);
result.Add(doc.GetField("creationDate").StringValue);
}
return result;
}
IndexWriter CreateIndex()
{
var directory = new RAMDirectory();
var analyzer = new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_30);
var writer = new IndexWriter(directory, analyzer, new IndexWriter.MaxFieldLength(1000));
return writer;
}
void Add(IndexWriter writer, IDictionary<string, string> values)
{
var document = new Document();
foreach (var kvp in values)
document.Add(new Field(kvp.Key, kvp.Value.ToLowerInvariant(), Field.Store.YES, Field.Index.ANALYZED));
document.Add(new NumericField("creationDate", Field.Store.YES, true).SetLongValue(DateTime.UtcNow.Ticks));
writer.AddDocument(document);
}
}
}