public void FullTextIncrementalIndexingLucene1()
{
//Lucene Index
Directory dir = new RAMDirectory();
LuceneSubjectsIndexer indexer = new LuceneSubjectsIndexer(dir, new StandardAnalyzer(LuceneTestHarness.LuceneVersion), new DefaultIndexSchema());
//Test Graph
Graph g = new Graph();
g.LoadFromEmbeddedResource("VDS.RDF.Configuration.configuration.ttl");
//Try indexing in 100 Triple chunks
Random rnd = new Random();
String searchTerm = "http";
for (int i = 0; i < g.Triples.Count; i += 100)
{
//Index the Triples
List <Triple> ts = g.Triples.Skip(i).Take(100).ToList();
foreach (Triple t in ts)
{
indexer.Index(t);
}
indexer.Flush();
//Now do a search to check some of those triples got indexed
//Pick the first multi-word string literal we can find from the batch and grab one word from it
INode targetNode = ts.Where(t => t.Object.NodeType == NodeType.Literal && t.Object.ToString().Contains("http")).Select(t => t.Subject).FirstOrDefault();
if (targetNode == null)
{
continue;
}
Console.WriteLine("Picked " + targetNode.ToString(this._formatter) + " as search target with search term '" + searchTerm + "'");
LuceneSearchProvider searcher = new LuceneSearchProvider(LuceneTestHarness.LuceneVersion, dir);
IEnumerable <IFullTextSearchResult> results = searcher.Match(searchTerm);
foreach (IFullTextSearchResult r in results)
{
Console.WriteLine("Got result " + r.Node.ToString(this._formatter) + " with score " + r.Score);
}
Assert.IsTrue(results.Any(r => r.Node.Equals(targetNode)), "Did not find expected node in search results");
searcher.Dispose();
Console.WriteLine();
}
indexer.Dispose();
}
public void FullTextDatasetLucene1()
{
//Lucene Index
Directory dir = new RAMDirectory();
LuceneSubjectsIndexer indexer = new LuceneSubjectsIndexer(dir, new StandardAnalyzer(LuceneTestHarness.LuceneVersion), new DefaultIndexSchema());
//Test Dataset
InMemoryDataset memData = new InMemoryDataset();
FullTextIndexedDataset dataset = new FullTextIndexedDataset(memData, indexer, false);
//Test Graph
Graph g = new Graph();
g.LoadFromEmbeddedResource("VDS.RDF.Configuration.configuration.ttl");
dataset.AddGraph(g);
Assert.True(dataset.HasGraph(g.BaseUri), "Graph should exist in dataset");
//Now do a search to check all the triples got indexed
String searchTerm = "http";
IEnumerable <Triple> searchTriples = g.Triples.Where(t => t.Object.NodeType == NodeType.Literal && t.Object.ToString().Contains("http"));
LuceneSearchProvider searcher = new LuceneSearchProvider(LuceneTestHarness.LuceneVersion, dir);
foreach (Triple searchTriple in searchTriples)
{
INode targetNode = searchTriple.Subject;
IEnumerable <IFullTextSearchResult> results = searcher.Match(searchTerm);
Assert.True(results.Any(r => r.Node.Equals(targetNode)), "Did not find expected node " + targetNode.ToString(this._formatter) + " in search results using search term '" + searchTerm + "' (found " + results.Count() + " results)");
Console.WriteLine();
}
//Now remove the Graph
dataset.RemoveGraph(g.BaseUri);
//Repeat the search to check all the triples got unindexed
foreach (Triple searchTriple in searchTriples)
{
INode targetNode = searchTriple.Subject;
IEnumerable <IFullTextSearchResult> results = searcher.Match(searchTerm);
Assert.False(results.Any(r => r.Node.Equals(targetNode)), "Found unexpected node " + targetNode.ToString(this._formatter) + " in search results using search term '" + searchTerm + "' (found " + results.Count() + " results)");
Console.WriteLine();
}
searcher.Dispose();
indexer.Dispose();
}
public void FullTextIndexCreationLuceneSubjects()
{
IFullTextIndexer indexer = null;
try
{
indexer = new LuceneSubjectsIndexer(LuceneTestHarness.Index, LuceneTestHarness.Analyzer, LuceneTestHarness.Schema);
indexer.Index(this.GetTestData());
}
finally
{
if (indexer != null)
{
indexer.Dispose();
}
}
}
public void FullTextConfigSerializeIndexerLuceneSubjects()
{
LuceneSubjectsIndexer indexer = new LuceneSubjectsIndexer(LuceneTestHarness.Index, LuceneTestHarness.Analyzer, LuceneTestHarness.Schema);
ConfigurationSerializationContext context = new ConfigurationSerializationContext();
INode obj = context.Graph.CreateBlankNode();
context.NextSubject = obj;
indexer.SerializeConfiguration(context);
indexer.Dispose();
TestTools.ShowGraph(context.Graph);
ConfigurationLoader.AutoConfigureObjectFactories(context.Graph);
Object temp = ConfigurationLoader.LoadObject(context.Graph, obj);
Assert.IsTrue(temp is LuceneSubjectsIndexer, "Should have returned a LuceneSubjectsIndexer instance");
Assert.IsTrue(temp is IFullTextIndexer, "Should have returned a IFullTextIndexer instance");
}
public void FullTextIndexSearchLuceneSubjectsWithLimitAndThreshold()
{
IFullTextIndexer indexer = null;
IFullTextSearchProvider provider = null;
try
{
indexer = new LuceneSubjectsIndexer(LuceneTestHarness.Index, LuceneTestHarness.Analyzer, LuceneTestHarness.Schema);
indexer.Index(this.GetTestData());
}
finally
{
if (indexer != null)
{
indexer.Dispose();
}
}
try
{
provider = new LuceneSearchProvider(LuceneTestHarness.LuceneVersion, LuceneTestHarness.Index);
NTriplesFormatter formatter = new NTriplesFormatter();
int i = 0;
foreach (IFullTextSearchResult result in provider.Match("http", 1.0d, 5))
{
i++;
Console.WriteLine(result.Node.ToString(formatter) + " - Scores " + result.Score);
Assert.IsTrue(result.Score >= 1.0d, "Score should be higher than desired threshold");
}
Assert.IsTrue(i <= 5, "Should be a max of 5 results");
}
finally
{
if (provider != null)
{
provider.Dispose();
}
}
}
public void FullTextIndexSearchLuceneSubjects()
{
IFullTextIndexer indexer = null;
IFullTextSearchProvider provider = null;
try
{
indexer = new LuceneSubjectsIndexer(LuceneTestHarness.Index, LuceneTestHarness.Analyzer, LuceneTestHarness.Schema);
indexer.Index(this.GetTestData());
}
finally
{
if (indexer != null)
{
indexer.Dispose();
}
}
try
{
provider = new LuceneSearchProvider(LuceneTestHarness.LuceneVersion, LuceneTestHarness.Index);
NTriplesFormatter formatter = new NTriplesFormatter();
foreach (IFullTextSearchResult result in provider.Match("http"))
{
Console.WriteLine(result.Node.ToString(formatter) + " - Scores " + result.Score);
}
}
finally
{
if (provider != null)
{
provider.Dispose();
}
}
}
public void FullTextIndexDestructionLuceneSubjects()
{
IFullTextIndexer indexer = null;
try
{
LuceneSearchProvider provider = null;
int origCount;
try
{
provider = new LuceneSearchProvider(LuceneTestHarness.LuceneVersion, LuceneTestHarness.Index);
origCount = provider.Match("http").Count();
provider.Dispose();
}
catch
{
origCount = 0;
}
finally
{
if (provider != null)
{
provider.Dispose();
}
}
Console.WriteLine("Prior to indexing search returns " + origCount + " result(s)");
indexer = new LuceneSubjectsIndexer(LuceneTestHarness.Index, LuceneTestHarness.Analyzer, LuceneTestHarness.Schema);
IGraph g = this.GetTestData();
indexer.Index(g);
indexer.Dispose();
indexer = null;
int currCount;
try
{
provider = new LuceneSearchProvider(LuceneTestHarness.LuceneVersion, LuceneTestHarness.Index);
currCount = provider.Match("http").Count();
provider.Dispose();
}
catch
{
currCount = 0;
}
finally
{
if (provider != null)
{
provider.Dispose();
}
}
Console.WriteLine("After indexing search returns " + currCount + " result(s)");
indexer = new LuceneSubjectsIndexer(LuceneTestHarness.Index, LuceneTestHarness.Analyzer, LuceneTestHarness.Schema);
indexer.Unindex(g);
indexer.Dispose();
indexer = null;
try
{
provider = new LuceneSearchProvider(LuceneTestHarness.LuceneVersion, LuceneTestHarness.Index);
currCount = provider.Match("http").Count();
Console.WriteLine("After unindexing search returns " + currCount + " result(s)");
}
finally
{
if (provider != null)
{
provider.Dispose();
}
}
Assert.Equal(origCount, currCount);
}
finally
{
if (indexer != null)
{
indexer.Dispose();
}
LuceneTestHarness.Index.Dispose();
}
}