public void IndexDocument(BaristaIndexDefinition indexDefinition, string documentId, DocumentDto document)
{
try
{
if (documentId.IsNullOrWhiteSpace())
{
throw new ArgumentNullException("documentId", @"A document id must be specified.");
}
if (document == null)
{
throw new ArgumentNullException("document", @"A document must be specified.");
}
var index = GetOrAddIndex(indexDefinition, true);
try
{
//Add it to the index.
var luceneDocument = DocumentDto.ConvertToLuceneDocument(document);
var batch = new IndexingBatch();
batch.Add(new BatchedDocument
{
DocumentId = documentId,
Document = luceneDocument,
SkipDeleteFromIndex = false,
});
index.IndexDocuments(batch);
}
catch (OutOfMemoryException)
{
CloseIndexWriter(indexDefinition, false);
}
}
catch (Exception ex)
{
throw new FaultException(ex.Message);
}
}
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);
}
public abstract void IndexDocuments(IndexingBatch batch);
public void IndexJsonDocuments(BaristaIndexDefinition indexDefinition, IEnumerable <JsonDocumentDto> documents)
{
try
{
if (documents == null)
{
throw new ArgumentNullException("documents", @"A collection of documents must be specified.");
}
var jsonDocuments = documents as IList <JsonDocumentDto> ?? documents.ToList();
if (jsonDocuments.Any() == false)
{
throw new ArgumentNullException("documents", @"At least one document must be contained within the collection.");
}
var index = GetOrAddIndex(indexDefinition, true);
try
{
//Add it to the index.
var batch = new IndexingBatch();
//Update the indexDefinition for the index based on the options specified.
foreach (var document in jsonDocuments)
{
UpdateIndexDefinitionFromFieldOptions(index.IndexDefinition, document.FieldOptions);
}
//Attempt to create a new Search.JsonDocument from the document
var searchJsonDocuments = jsonDocuments.Select(document => new Search.JsonDocument
{
DocumentId = document.DocumentId,
Metadata = document.MetadataAsJson.IsNullOrWhiteSpace() == false
? JObject.Parse(document.MetadataAsJson)
: new JObject(),
DataAsJson = JObject.Parse(document.DataAsJson)
});
var luceneDocuments =
JsonDocumentToLuceneDocumentConverter.ConvertJsonDocumentToLuceneDocument(index.IndexDefinition,
searchJsonDocuments);
foreach (var luceneDocument in luceneDocuments)
{
batch.Add(luceneDocument);
}
//TODO: Add the batch to a BlockingCollection<IndexingBatch> and run a thread that consumes the batches
//See http://www.codethinked.com/blockingcollection-and-iproducerconsumercollection
index.IndexDocuments(batch);
}
catch (OutOfMemoryException)
{
CloseIndexWriter(indexDefinition, false);
}
}
catch (Exception ex)
{
throw new FaultException(ex.Message);
}
}
