/// <summary> Check whether the word exists in the index.</summary>
/// <param name="word">String
/// </param>
/// <throws> IOException </throws>
/// <returns> true iff the word exists in the index
/// </returns>
public virtual bool Exist(System.String word)
{
// obtainSearcher calls ensureOpen
IndexSearcher indexSearcher = ObtainSearcher();
try
{
return(indexSearcher.DocFreq(F_WORD_TERM.CreateTerm(word)) > 0);
}
finally
{
ReleaseSearcher(indexSearcher);
}
}
public override Query Rewrite(IndexReader reader)
{
if (_termIsPrefix)
{
MultiTermQuery rewritten =
new PrefixQuery(internalTerm.CreateTerm(internalTerm.Text.Substring(0, internalTerm.Text.IndexOf('*'))));
rewritten.Boost = Boost;
rewritten.RewriteMethod = RewriteMethod;
return(rewritten);
}
else
{
return(base.Rewrite(reader));
}
}
public override void IndexDocuments(IndexingBatch batch)
{
var count = 0;
var sourceCount = 0;
var sw = Stopwatch.StartNew();
var start = DateTime.UtcNow;
Write((indexWriter, analyzer) =>
{
//TODO: The following would be a perfect candidate for a TPL DataFlow impl. Too bad we're currently on .Net 3.5
var processedKeys = new HashSet <string>();
var docIdTerm = new Lucene.Net.Index.Term(Constants.DocumentIdFieldName);
var documentsWrapped = batch.Documents.Select((doc, i) =>
{
Interlocked.Increment(ref sourceCount);
if (doc.DocumentId == null)
{
throw new ArgumentException(
string.Format("Cannot index something which doesn't have a document id, but got: '{0}'", doc));
}
var documentId = doc.DocumentId.ToString(CultureInfo.InvariantCulture);
if (processedKeys.Add(documentId) == false)
{
return(doc);
}
if (doc.SkipDeleteFromIndex == false)
{
indexWriter.DeleteDocuments(docIdTerm.CreateTerm(documentId.ToLowerInvariant()));
}
return(doc);
})
.ToList();
foreach (var document in documentsWrapped)
{
Interlocked.Increment(ref count);
LogIndexedDocument(document.DocumentId, document.Document);
AddDocumentToIndex(indexWriter, document.Document, analyzer);
indexWriter.Commit();
}
return(sourceCount);
});
AddindexingPerformanceStat(new IndexingPerformanceStats
{
OutputCount = count,
InputCount = sourceCount,
Duration = sw.Elapsed,
Operation = "Index",
Started = start
});
LogIndexing.Debug("Indexed {0} documents for {1}", count, Name);
}
