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 FullTextGraphScopingTests()
{
String data = @"<http://x> <http://p> ""This is sample text"" <http://g1> .
<http://y> <http://p> ""This is sample text"" <http://g2> .
<http://y> <http://p> ""Additional sample"" <http://g2> .";
this._store = new TripleStore();
StringParser.ParseDataset(this._store, data, new NQuadsParser());
this._index = new RAMDirectory();
using (LuceneSubjectsIndexer indexer = new LuceneSubjectsIndexer(this._index, new StandardAnalyzer(LucUtil.Version.LUCENE_30), new DefaultIndexSchema()))
{
foreach (IGraph g in this._store.Graphs)
{
indexer.Index(g);
}
}
}
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();
}
}
}
/// <summary>
/// Tries to load an object based on information from the Configuration Graph
/// </summary>
/// <param name="g">Configuration Graph</param>
/// <param name="objNode">Object Node</param>
/// <param name="targetType">Target Type</param>
/// <param name="obj">Output Object</param>
/// <returns></returns>
public bool TryLoadObject(IGraph g, INode objNode, Type targetType, out object obj)
{
obj = null;
INode index = g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "index"));
//INode indexer = g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "indexer"));
INode searcher = g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "searcher"));
INode analyzer = g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "analyzer"));
INode schema = g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "schema"));
INode version = g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "version"));
Object tempIndex, tempAnalyzer, tempSchema;
int ver = 2900;
//Always check for the version
ver = ConfigurationLoader.GetConfigurationInt32(g, objNode, version, 2900);
switch (targetType.FullName)
{
case DefaultIndexSchema:
obj = new DefaultIndexSchema();
break;
case FullTextOptimiser:
//Need to get the Search Provider
INode providerNode = ConfigurationLoader.GetConfigurationNode(g, objNode, searcher);
if (providerNode == null)
{
throw new DotNetRdfConfigurationException("Unable to load the Full Text Optimiser specified by the Node '" + objNode.ToString() + "' as there was no value specified for the required dnr-ft:searcher property");
}
Object tempSearcher = ConfigurationLoader.LoadObject(g, providerNode);
if (tempSearcher is IFullTextSearchProvider)
{
obj = new FullTextOptimiser((IFullTextSearchProvider)tempSearcher);
}
else
{
throw new DotNetRdfConfigurationException("Unable to load the Full Text Optimiser specified by the Node '" + objNode.ToString() + "' as the value specified for the dnr-ft:searcher property pointed to an object which could not be loaded as a type that implements the required IFullTextSearchProvider interface");
}
break;
case LuceneObjectsIndexer:
case LucenePredicatesIndexer:
case LuceneSubjectsIndexer:
case LuceneSearchProvider:
//For any Lucene Indexer/Search Provider need to know the Index, Analyzer and Schema to be used
//Then get the Index
tempIndex = ConfigurationLoader.GetConfigurationNode(g, objNode, index);
if (tempIndex == null)
{
throw new DotNetRdfConfigurationException("Unable to load the Lucene Indexer specified by the Node '" + objNode.ToString() + "' as there was no value specified for the required dnr-ft:index property");
}
tempIndex = ConfigurationLoader.LoadObject(g, (INode)tempIndex);
if (tempIndex is Directory)
{
//Next get the Analyzer (assume Standard if none specified)
tempAnalyzer = ConfigurationLoader.GetConfigurationNode(g, objNode, analyzer);
if (tempAnalyzer == null)
{
tempAnalyzer = new StandardAnalyzer(this.GetLuceneVersion(ver));
}
else
{
tempAnalyzer = ConfigurationLoader.LoadObject(g, (INode)tempAnalyzer);
}
if (tempAnalyzer is Analyzer)
{
//Finally get the Schema (assume Default if none specified)
tempSchema = ConfigurationLoader.GetConfigurationNode(g, objNode, schema);
if (tempSchema == null)
{
tempSchema = new DefaultIndexSchema();
}
else
{
tempSchema = ConfigurationLoader.LoadObject(g, (INode)tempSchema);
}
if (tempSchema is IFullTextIndexSchema)
{
//Now we can create the Object
switch (targetType.FullName)
{
case LuceneObjectsIndexer:
obj = new LuceneObjectsIndexer((Directory)tempIndex, (Analyzer)tempAnalyzer, (IFullTextIndexSchema)tempSchema);
break;
case LucenePredicatesIndexer:
obj = new LucenePredicatesIndexer((Directory)tempIndex, (Analyzer)tempAnalyzer, (IFullTextIndexSchema)tempSchema);
break;
case LuceneSubjectsIndexer:
obj = new LuceneSubjectsIndexer((Directory)tempIndex, (Analyzer)tempAnalyzer, (IFullTextIndexSchema)tempSchema);
break;
case LuceneSearchProvider:
//Before the Search Provider has been loaded determine whether we need to carry out auto-indexing
List <INode> sources = ConfigurationLoader.GetConfigurationData(g, objNode, g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "buildIndexFor"))).ToList();
if (sources.Count > 0)
{
//If there are sources to index ensure we have an indexer to index with
INode indexerNode = ConfigurationLoader.GetConfigurationNode(g, objNode, g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "buildIndexWith")));
if (indexerNode == null)
{
throw new DotNetRdfConfigurationException("Unable to load the Lucene Search Provider specified by the Node '" + objNode.ToString() + "' as there were values specified for the dnr-ft:buildIndexFor property but no dnr-ft:buildIndexWith property was found");
}
IFullTextIndexer indexer = ConfigurationLoader.LoadObject(g, indexerNode) as IFullTextIndexer;
if (indexer == null)
{
throw new DotNetRdfConfigurationException("Unable to load the Lucene Search Provider specified by the Node '" + objNode.ToString() + "' as the value given for the dnr-ft:buildIndexWith property pointed to an Object which could not be loaded as a type that implements the required IFullTextIndexer interface");
}
try
{
//For Each Source load it and Index it
foreach (INode sourceNode in sources)
{
Object source = ConfigurationLoader.LoadObject(g, sourceNode);
if (source is ISparqlDataset)
{
indexer.Index((ISparqlDataset)source);
}
else if (source is ITripleStore)
{
foreach (IGraph graph in ((ITripleStore)source).Graphs)
{
indexer.Index(graph);
}
}
else if (source is IGraph)
{
indexer.Index((IGraph)source);
}
else
{
throw new DotNetRdfConfigurationException("Unable to load the Lucene Search Provider specified by the Node '" + objNode.ToString() + "' as a value given for the dnr-ft:buildIndexFor property ('" + sourceNode.ToString() + "') pointed to an Object which could not be loaded as a type that implements one of the required interfaces: IGraph, ITripleStore or ISparqlDataset");
}
}
}
finally
{
indexer.Dispose();
}
}
//Then we actually load the Search Provider
obj = new LuceneSearchProvider(this.GetLuceneVersion(ver), (Directory)tempIndex, (Analyzer)tempAnalyzer, (IFullTextIndexSchema)tempSchema);
break;
}
}
else
{
throw new DotNetRdfConfigurationException("Unable to load the Lucene Indexer specified by the Node '" + objNode.ToString() + "' as the value given for the dnr-ft:schema property pointed to an Object which could not be loaded as a type that implements the required IFullTextIndexSchema interface");
}
}
else
{
throw new DotNetRdfConfigurationException("Unable to load the Lucene Indexer specified by the Node '" + objNode.ToString() + "' as the value given for the dnr-ft:analyzer property pointed to an Object which could not be loaded as a type that derives from the required Lucene.Net.Analysis.Analyzer type");
}
}
else
{
throw new DotNetRdfConfigurationException("Unable to load the Lucene Indexer specified by the Node '" + objNode.ToString() + "' as the value given for the dnr-ft:index property pointed to an Object which could not be loaded as a type that derives from the required Lucene.Net.Store.Directory type");
}
break;
default:
try
{
if (this._luceneAnalyzerType.IsAssignableFrom(targetType))
{
if (targetType.GetConstructor(new Type[] { typeof(LucVersion) }) != null)
{
obj = Activator.CreateInstance(targetType, new Object[] { this.GetLuceneVersion(ver) });
}
else
{
obj = Activator.CreateInstance(targetType);
}
}
else if (this._luceneDirectoryType.IsAssignableFrom(targetType))
{
String dir = ConfigurationLoader.GetConfigurationString(g, objNode, ConfigurationLoader.CreateConfigurationNode(g, ConfigurationLoader.PropertyFromFile));
if (dir != null)
{
try
{
obj = Activator.CreateInstance(targetType, new Object[] { dir });
}
catch
{
MethodInfo method = targetType.GetMethod("Open", new Type[] { typeof(DirInfo) });
if (method != null)
{
obj = method.Invoke(null, new Object[] { new DirInfo(ConfigurationLoader.ResolvePath(dir)) });
}
}
}
else
{
obj = Activator.CreateInstance(targetType);
}
//Ensure the Index if necessary
if (obj != null)
{
if (ConfigurationLoader.GetConfigurationBoolean(g, objNode, g.CreateUriNode(new Uri(FullTextHelper.FullTextConfigurationNamespace + "ensureIndex")), false))
{
IndexWriter writer = new IndexWriter((Directory)obj, new StandardAnalyzer(this.GetLuceneVersion(ver)));
writer.Close();
}
}
}
}
catch
{
//Since we know we don't allow loading of all analyzers and directories we allow for users to inject other object factories
//which may know how to load those specific instances
obj = null;
}
break;
}
return(obj != null);
}
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();
}
}
