public virtual void TestRequireDimCount()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, dir);
DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, OpenMode.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetRequireDimCount("dim", true);
config.SetMultiValued("dim2", true);
config.SetRequireDimCount("dim2", true);
config.SetMultiValued("dim3", true);
config.SetHierarchical("dim3", true);
config.SetRequireDimCount("dim3", true);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
doc.Add(new FacetField("dim", "a"));
doc.Add(new FacetField("dim2", "a"));
doc.Add(new FacetField("dim2", "b"));
doc.Add(new FacetField("dim3", "a", "b"));
doc.Add(new FacetField("dim3", "a", "c"));
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.GetReader());
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
FacetsCollector c = new FacetsCollector();
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim").Value);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim2").Value);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim3").Value);
try
{
Assert.AreEqual(1, facets.GetSpecificValue("dim"));
fail("didn't hit expected exception");
}
catch (Exception iae) when(iae.IsIllegalArgumentException())
{
// expected
}
Assert.AreEqual(1, facets.GetSpecificValue("dim2"));
Assert.AreEqual(1, facets.GetSpecificValue("dim3"));
IOUtils.Dispose(writer, taxoWriter, searcher.IndexReader, taxoReader, dir, taxoDir);
}
/// <summary>Empty constructor</summary>
public AssociationsFacetsExample()
{
config = new FacetsConfig();
config.SetMultiValued("tags", true);
config.SetIndexFieldName("tags", "$tags");
config.SetMultiValued("genre", true);
config.SetIndexFieldName("genre", "$genre");
}
public virtual void TestRequireDimCount()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, Similarity, TimeZone);
DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode_e.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetRequireDimCount("dim", true);
config.SetMultiValued("dim2", true);
config.SetRequireDimCount("dim2", true);
config.SetMultiValued("dim3", true);
config.SetHierarchical("dim3", true);
config.SetRequireDimCount("dim3", true);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
doc.Add(new FacetField("dim", "a"));
doc.Add(new FacetField("dim2", "a"));
doc.Add(new FacetField("dim2", "b"));
doc.Add(new FacetField("dim3", "a", "b"));
doc.Add(new FacetField("dim3", "a", "c"));
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.Reader);
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
FacetsCollector c = new FacetsCollector();
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim").Value);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim2").Value);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim3").Value);
try
{
Assert.AreEqual(1, facets.GetSpecificValue("dim"));
Fail("didn't hit expected exception");
}
catch (System.ArgumentException)
{
// expected
}
Assert.AreEqual(1, facets.GetSpecificValue("dim2"));
Assert.AreEqual(1, facets.GetSpecificValue("dim3"));
IOUtils.Close(writer, taxoWriter, searcher.IndexReader, taxoReader, dir, taxoDir);
}
private static FacetsConfig GetConfig()
{
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("A", true);
config.SetMultiValued("B", true);
config.SetRequireDimCount("B", true);
config.SetHierarchical("D", true);
return(config);
}
public override void BeforeClass()
{
base.BeforeClass();
dir = NewDirectory();
taxoDir = NewDirectory();
// preparations - index, taxonomy, content
var taxoWriter = new DirectoryTaxonomyWriter(taxoDir);
// Cannot mix ints & floats in the same indexed field:
config = new FacetsConfig();
config.SetIndexFieldName("int", "$facets.int");
config.SetMultiValued("int", true);
config.SetIndexFieldName("float", "$facets.float");
config.SetMultiValued("float", true);
var writer = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, dir);
// index documents, 50% have only 'b' and all have 'a'
for (int i = 0; i < 110; i++)
{
Document doc = new Document();
// every 11th document is added empty, this used to cause the association
// aggregators to go into an infinite loop
if (i % 11 != 0)
{
doc.Add(new Int32AssociationFacetField(2, "int", "a"));
doc.Add(new SingleAssociationFacetField(0.5f, "float", "a"));
if (i % 2 == 0) // 50
{
doc.Add(new Int32AssociationFacetField(3, "int", "b"));
doc.Add(new SingleAssociationFacetField(0.2f, "float", "b"));
}
}
writer.AddDocument(config.Build(taxoWriter, doc));
}
taxoWriter.Dispose();
reader = writer.GetReader();
writer.Dispose();
taxoReader = new DirectoryTaxonomyReader(taxoDir);
}
/// <summary>
/// Add documents.
/// </summary>
/// <param name="writer">The index writer.</param>
/// <param name="facetWriter">The facet index writer.</param>
/// <param name="facetData">The complete facet information used to build the index information.</param>
public void AddDocuments(Lucene.Net.Index.IndexWriter writer, DirectoryTaxonomyWriter facetWriter, FacetData facetData)
{
// Build the facet configuration information.
FacetsConfig config = new FacetsConfig();
// Builder hierarchicals.
if (facetData.Hierarchicals != null && facetData.Hierarchicals.Length > 0)
{
// Add the config.
foreach (FacetData.Hierarchical item in facetData.Hierarchicals)
{
config.SetHierarchical(item.DimensionName, item.IsHierarchical);
}
}
// Builder index fields.
if (facetData.IndexFields != null && facetData.IndexFields.Length > 0)
{
// Add the config.
foreach (FacetData.IndexField item in facetData.IndexFields)
{
config.SetIndexFieldName(item.DimensionName, item.IndexFieldName);
}
}
// Builder multi values.
if (facetData.MultiValues != null && facetData.MultiValues.Length > 0)
{
// Add the config.
foreach (FacetData.MultiValued item in facetData.MultiValues)
{
config.SetMultiValued(item.DimensionName, item.IsMultiValue);
}
}
// Builder require dimension counts.
if (facetData.RequireDimensionCounts != null && facetData.RequireDimensionCounts.Length > 0)
{
// Add the config.
foreach (FacetData.RequireDimensionCount item in facetData.RequireDimensionCounts)
{
config.SetRequireDimCount(item.DimensionName, item.IsAccurateCountsRequired);
}
}
// Add text data.
if (facetData.TextFacetFields.Count > 0)
{
// Add the text.
AddText(writer, facetWriter, facetData.TextFacetFields, config);
}
// Add file data.
if (facetData.FileFacetFields.Count > 0)
{
// Add the file.
AddFile(writer, facetWriter, facetData.FileFacetFields, config);
}
}
public override void Configure(FacetsConfig config)
{
for (int i = 0; i < maxDims; i++)
{
config.SetHierarchical(i.ToString(CultureInfo.InvariantCulture), true);
config.SetMultiValued(i.ToString(CultureInfo.InvariantCulture), true);
}
}
public virtual void TestBasic()
{
AssumeTrue("Test requires SortedSetDV support", DefaultCodecSupportsSortedSet);
Directory dir = NewDirectory();
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("a", true);
RandomIndexWriter writer = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, dir);
Document doc = new Document();
doc.Add(new SortedSetDocValuesFacetField("a", "foo"));
doc.Add(new SortedSetDocValuesFacetField("a", "bar"));
doc.Add(new SortedSetDocValuesFacetField("a", "zoo"));
doc.Add(new SortedSetDocValuesFacetField("b", "baz"));
writer.AddDocument(config.Build(doc));
if (Random.NextBoolean())
{
writer.Commit();
}
doc = new Document();
doc.Add(new SortedSetDocValuesFacetField("a", "foo"));
writer.AddDocument(config.Build(doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.GetReader());
// Per-top-reader state:
SortedSetDocValuesReaderState state = new DefaultSortedSetDocValuesReaderState(searcher.IndexReader);
FacetsCollector c = new FacetsCollector();
searcher.Search(new MatchAllDocsQuery(), c);
SortedSetDocValuesFacetCounts facets = new SortedSetDocValuesFacetCounts(state, c);
Assert.AreEqual("dim=a path=[] value=4 childCount=3\n foo (2)\n bar (1)\n zoo (1)\n", facets.GetTopChildren(10, "a").ToString());
Assert.AreEqual("dim=b path=[] value=1 childCount=1\n baz (1)\n", facets.GetTopChildren(10, "b").ToString());
// DrillDown:
DrillDownQuery q = new DrillDownQuery(config);
q.Add("a", "foo");
q.Add("b", "baz");
TopDocs hits = searcher.Search(q, 1);
Assert.AreEqual(1, hits.TotalHits);
IOUtils.Dispose(writer, searcher.IndexReader, dir);
}
public virtual void TestMultiValuedHierarchy()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, OpenMode.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetHierarchical("a", true);
config.SetMultiValued("a", true);
RandomIndexWriter writer = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, dir);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
doc.Add(new FacetField("a", "path", "x"));
doc.Add(new FacetField("a", "path", "y"));
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.GetReader());
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
// Aggregate the facet counts:
FacetsCollector c = new FacetsCollector();
// MatchAllDocsQuery is for "browsing" (counts facets
// for all non-deleted docs in the index); normally
// you'd use a "normal" query, and use MultiCollector to
// wrap collecting the "normal" hits and also facets:
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
try
{
facets.GetSpecificValue("a");
fail("didn't hit expected exception");
}
catch (Exception iae) when(iae.IsIllegalArgumentException())
{
// expected
}
FacetResult result = facets.GetTopChildren(10, "a");
Assert.AreEqual(1, result.LabelValues.Length);
Assert.AreEqual(1, (int)result.LabelValues[0].Value);
IOUtils.Dispose(writer, taxoWriter, searcher.IndexReader, taxoReader, dir, taxoDir);
}
public virtual void TestManyFacetsInOneDocument()
{
AssumeTrue("default Codec doesn't support huge BinaryDocValues", TestUtil.FieldSupportsHugeBinaryDocValues(FacetsConfig.DEFAULT_INDEX_FIELD_NAME));
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, iwc);
var taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode_e.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("dim", true);
int numLabels = TestUtil.NextInt(Random(), 40000, 100000);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
for (int i = 0; i < numLabels; i++)
{
doc.Add(new FacetField("dim", "" + i));
}
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.Reader);
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
// Aggregate the facet counts:
FacetsCollector c = new FacetsCollector();
// MatchAllDocsQuery is for "browsing" (counts facets
// for all non-deleted docs in the index); normally
// you'd use a "normal" query, and use MultiCollector to
// wrap collecting the "normal" hits and also facets:
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
FacetResult result = facets.GetTopChildren(int.MaxValue, "dim");
Assert.AreEqual(numLabels, result.LabelValues.Length);
var allLabels = new HashSet <string>();
foreach (LabelAndValue labelValue in result.LabelValues)
{
allLabels.Add(labelValue.Label);
Assert.AreEqual(1, (int)labelValue.Value);
}
Assert.AreEqual(numLabels, allLabels.Count);
IOUtils.Close(searcher.IndexReader, taxoWriter, writer, taxoReader, dir, taxoDir);
}
public void BeforeClass()
{
dir = NewDirectory();
taxoDir = NewDirectory();
// preparations - index, taxonomy, content
var taxoWriter = new DirectoryTaxonomyWriter(taxoDir);
// Cannot mix ints & floats in the same indexed field:
config = new FacetsConfig();
config.SetIndexFieldName("int", "$facets.int");
config.SetMultiValued("int", true);
config.SetIndexFieldName("float", "$facets.float");
config.SetMultiValued("float", true);
var writer = new RandomIndexWriter(Random(), dir, Similarity, TimeZone);
// index documents, 50% have only 'b' and all have 'a'
for (int i = 0; i < 110; i++)
{
Document doc = new Document();
// every 11th document is added empty, this used to cause the association
// aggregators to go into an infinite loop
if (i % 11 != 0)
{
doc.Add(new IntAssociationFacetField(2, "int", "a"));
doc.Add(new FloatAssociationFacetField(0.5f, "float", "a"));
if (i % 2 == 0) // 50
{
doc.Add(new IntAssociationFacetField(3, "int", "b"));
doc.Add(new FloatAssociationFacetField(0.2f, "float", "b"));
}
}
writer.AddDocument(config.Build(taxoWriter, doc));
}
taxoWriter.Dispose();
reader = writer.Reader;
writer.Dispose();
taxoReader = new DirectoryTaxonomyReader(taxoDir);
}
public static void BeforeClass()
{
dir = NewDirectory();
taxoDir = NewDirectory();
// preparations - index, taxonomy, content
var taxoWriter = new DirectoryTaxonomyWriter(taxoDir);
// Cannot mix ints & floats in the same indexed field:
config = new FacetsConfig();
config.SetIndexFieldName("int", "$facets.int");
config.SetMultiValued("int", true);
config.SetIndexFieldName("float", "$facets.float");
config.SetMultiValued("float", true);
var writer = new RandomIndexWriter(Random(), dir);
// index documents, 50% have only 'b' and all have 'a'
for (int i = 0; i < 110; i++)
{
Document doc = new Document();
// every 11th document is added empty, this used to cause the association
// aggregators to go into an infinite loop
if (i % 11 != 0)
{
doc.Add(new IntAssociationFacetField(2, "int", "a"));
doc.Add(new FloatAssociationFacetField(0.5f, "float", "a"));
if (i % 2 == 0) // 50
{
doc.Add(new IntAssociationFacetField(3, "int", "b"));
doc.Add(new FloatAssociationFacetField(0.2f, "float", "b"));
}
}
writer.AddDocument(config.Build(taxoWriter, doc));
}
taxoWriter.Dispose();
reader = writer.Reader;
writer.Dispose();
taxoReader = new DirectoryTaxonomyReader(taxoDir);
}
public virtual void TestBasic()
{
AssumeTrue("Test requires SortedSetDV support", DefaultCodecSupportsSortedSet());
Directory dir = NewDirectory();
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("a", true);
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, Similarity, TimeZone);
Document doc = new Document();
doc.Add(new SortedSetDocValuesFacetField("a", "foo"));
doc.Add(new SortedSetDocValuesFacetField("a", "bar"));
doc.Add(new SortedSetDocValuesFacetField("a", "zoo"));
doc.Add(new SortedSetDocValuesFacetField("b", "baz"));
writer.AddDocument(config.Build(doc));
if (Random().NextBoolean())
{
writer.Commit();
}
doc = new Document();
doc.Add(new SortedSetDocValuesFacetField("a", "foo"));
writer.AddDocument(config.Build(doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.Reader);
// Per-top-reader state:
SortedSetDocValuesReaderState state = new DefaultSortedSetDocValuesReaderState(searcher.IndexReader);
FacetsCollector c = new FacetsCollector();
searcher.Search(new MatchAllDocsQuery(), c);
SortedSetDocValuesFacetCounts facets = new SortedSetDocValuesFacetCounts(state, c);
Assert.AreEqual("dim=a path=[] value=4 childCount=3\n foo (2)\n bar (1)\n zoo (1)\n", facets.GetTopChildren(10, "a").ToString());
Assert.AreEqual("dim=b path=[] value=1 childCount=1\n baz (1)\n", facets.GetTopChildren(10, "b").ToString());
// DrillDown:
DrillDownQuery q = new DrillDownQuery(config);
q.Add("a", "foo");
q.Add("b", "baz");
TopDocs hits = searcher.Search(q, 1);
Assert.AreEqual(1, hits.TotalHits);
IOUtils.Close(writer, searcher.IndexReader, dir);
}
public virtual void TestLabelWithDelimiter()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, dir);
var taxoWriter = new DirectoryTaxonomyWriter(taxoDir, OpenMode.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("dim", true);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
doc.Add(new FacetField("dim", "test\u001Fone"));
doc.Add(new FacetField("dim", "test\u001Etwo"));
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.GetReader());
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
FacetsCollector c = new FacetsCollector();
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
Assert.AreEqual(1, facets.GetSpecificValue("dim", "test\u001Fone"));
Assert.AreEqual(1, facets.GetSpecificValue("dim", "test\u001Etwo"));
FacetResult result = facets.GetTopChildren(10, "dim");
Assert.AreEqual("dim=dim path=[] value=-1 childCount=2\n test\u001Fone (1)\n test\u001Etwo (1)\n", result.ToString());
IOUtils.Dispose(writer, taxoWriter, searcher.IndexReader, taxoReader, dir, taxoDir);
}
public virtual void TestManyFacetsInOneDocument()
{
AssumeTrue("default Codec doesn't support huge BinaryDocValues", TestUtil.FieldSupportsHugeBinaryDocValues(FacetsConfig.DEFAULT_INDEX_FIELD_NAME));
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, iwc);
var taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode_e.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("dim", true);
int numLabels = TestUtil.NextInt(Random(), 40000, 100000);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
for (int i = 0; i < numLabels; i++)
{
doc.Add(new FacetField("dim", "" + i));
}
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.Reader);
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
// Aggregate the facet counts:
FacetsCollector c = new FacetsCollector();
// MatchAllDocsQuery is for "browsing" (counts facets
// for all non-deleted docs in the index); normally
// you'd use a "normal" query, and use MultiCollector to
// wrap collecting the "normal" hits and also facets:
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
FacetResult result = facets.GetTopChildren(int.MaxValue, "dim");
Assert.AreEqual(numLabels, result.LabelValues.Length);
var allLabels = new HashSet<string>();
foreach (LabelAndValue labelValue in result.LabelValues)
{
allLabels.Add(labelValue.label);
Assert.AreEqual(1, (int)labelValue.value);
}
Assert.AreEqual(numLabels, allLabels.Count);
IOUtils.Close(searcher.IndexReader, taxoWriter, writer, taxoReader, dir, taxoDir);
}
public virtual void TestConcurrency()
{
AtomicInt32 numDocs = new AtomicInt32(AtLeast(10000));
Directory indexDir = NewDirectory();
Directory taxoDir = NewDirectory();
ConcurrentDictionary <string, string> values = new ConcurrentDictionary <string, string>();
IndexWriter iw = new IndexWriter(indexDir, NewIndexWriterConfig(TEST_VERSION_CURRENT, null));
var tw = new DirectoryTaxonomyWriter(taxoDir, OpenMode.CREATE, NewTaxoWriterCache(numDocs));
ThreadJob[] indexThreads = new ThreadJob[AtLeast(4)];
FacetsConfig config = new FacetsConfig();
for (int i = 0; i < 10; i++)
{
config.SetHierarchical("l1." + i, true);
config.SetMultiValued("l1." + i, true);
}
for (int i = 0; i < indexThreads.Length; i++)
{
indexThreads[i] = new ThreadAnonymousInnerClassHelper(this, numDocs, values, iw, tw, config);
}
foreach (ThreadJob t in indexThreads)
{
t.Start();
}
foreach (ThreadJob t in indexThreads)
{
t.Join();
}
var tr = new DirectoryTaxonomyReader(tw);
// +1 for root category
if (values.Count + 1 != tr.Count)
{
foreach (string value in values.Keys)
{
FacetLabel label = new FacetLabel(FacetsConfig.StringToPath(value));
if (tr.GetOrdinal(label) == -1)
{
Console.WriteLine("FAIL: path=" + label + " not recognized");
}
}
fail("mismatch number of categories");
}
int[] parents = tr.ParallelTaxonomyArrays.Parents;
foreach (string cat in values.Keys)
{
FacetLabel cp = new FacetLabel(FacetsConfig.StringToPath(cat));
Assert.IsTrue(tr.GetOrdinal(cp) > 0, "category not found " + cp);
int level = cp.Length;
int parentOrd = 0; // for root, parent is always virtual ROOT (ord=0)
FacetLabel path = null;
for (int i = 0; i < level; i++)
{
path = cp.Subpath(i + 1);
int ord = tr.GetOrdinal(path);
Assert.AreEqual(parentOrd, parents[ord], "invalid parent for cp=" + path);
parentOrd = ord; // next level should have this parent
}
}
IOUtils.Dispose(tw, iw, tr, taxoDir, indexDir);
}
public virtual void Test_Directory() // LUCENENET specific - name collides with property of LuceneTestCase
{
Store.Directory indexDir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
IndexWriter w = new IndexWriter(indexDir, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random)));
var tw = new DirectoryTaxonomyWriter(taxoDir);
// first empty commit
w.Commit();
tw.Commit();
var mgr = new SearcherTaxonomyManager(indexDir, taxoDir, null);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("field", true);
AtomicBoolean stop = new AtomicBoolean();
// How many unique facets to index before stopping:
//int ordLimit = TestNightly ? 100000 : 6000;
// LUCENENET specific: 100000 facets takes about 2-3 hours. To keep it under
// the 1 hour free limit of Azure DevOps, this was reduced to 30000.
int ordLimit = TestNightly ? 30000 : 6000;
var indexer = new IndexerThread(w, config, tw, mgr, ordLimit, stop);
indexer.Start();
try
{
while (!stop)
{
SearcherAndTaxonomy pair = mgr.Acquire();
try
{
//System.out.println("search maxOrd=" + pair.taxonomyReader.getSize());
FacetsCollector sfc = new FacetsCollector();
pair.Searcher.Search(new MatchAllDocsQuery(), sfc);
Facets facets = GetTaxonomyFacetCounts(pair.TaxonomyReader, config, sfc);
FacetResult result = facets.GetTopChildren(10, "field");
if (pair.Searcher.IndexReader.NumDocs > 0)
{
//System.out.println(pair.taxonomyReader.getSize());
Assert.IsTrue(result.ChildCount > 0);
Assert.IsTrue(result.LabelValues.Length > 0);
}
//if (VERBOSE) {
//System.out.println("TEST: facets=" + FacetTestUtils.toString(results.get(0)));
//}
}
finally
{
mgr.Release(pair);
}
}
}
finally
{
indexer.Join();
}
if (Verbose)
{
Console.WriteLine("TEST: now stop");
}
IOUtils.Dispose(mgr, tw, w, taxoDir, indexDir);
}
public virtual void TestNrt()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
// Don't allow tiny maxBufferedDocs; it can make this
// test too slow:
iwc.SetMaxBufferedDocs(Math.Max(500, iwc.MaxBufferedDocs));
// MockRandom/AlcololicMergePolicy are too slow:
TieredMergePolicy tmp = new TieredMergePolicy();
tmp.FloorSegmentMB = .001;
iwc.SetMergePolicy(tmp);
IndexWriter w = new IndexWriter(dir, iwc);
var tw = new DirectoryTaxonomyWriter(taxoDir);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("field", true);
AtomicBoolean stop = new AtomicBoolean();
// How many unique facets to index before stopping:
int ordLimit = TEST_NIGHTLY ? 100000 : 6000;
var indexer = new IndexerThread(w, config, tw, null, ordLimit, stop);
var mgr = new SearcherTaxonomyManager(w, true, null, tw);
var reopener = new ThreadAnonymousInnerClassHelper(this, stop, mgr);
reopener.Name = "reopener";
reopener.Start();
indexer.Name = "indexer";
indexer.Start();
try
{
while (!stop.Get())
{
SearcherAndTaxonomy pair = mgr.Acquire();
try
{
//System.out.println("search maxOrd=" + pair.taxonomyReader.getSize());
FacetsCollector sfc = new FacetsCollector();
pair.searcher.Search(new MatchAllDocsQuery(), sfc);
Facets facets = GetTaxonomyFacetCounts(pair.taxonomyReader, config, sfc);
FacetResult result = facets.GetTopChildren(10, "field");
if (pair.searcher.IndexReader.NumDocs > 0)
{
//System.out.println(pair.taxonomyReader.getSize());
Assert.True(result.ChildCount > 0);
Assert.True(result.LabelValues.Length > 0);
}
//if (VERBOSE) {
//System.out.println("TEST: facets=" + FacetTestUtils.toString(results.get(0)));
//}
}
finally
{
mgr.Release(pair);
}
}
}
finally
{
indexer.Join();
reopener.Join();
}
if (VERBOSE)
{
Console.WriteLine("TEST: now stop");
}
IOUtils.Close(mgr, tw, w, taxoDir, dir);
}
public virtual void TestDirectory()
{
Store.Directory indexDir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
IndexWriter w = new IndexWriter(indexDir, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())));
var tw = new DirectoryTaxonomyWriter(taxoDir);
// first empty commit
w.Commit();
tw.Commit();
var mgr = new SearcherTaxonomyManager(indexDir, taxoDir, null);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("field", true);
AtomicBoolean stop = new AtomicBoolean();
// How many unique facets to index before stopping:
int ordLimit = TEST_NIGHTLY ? 100000 : 6000;
var indexer = new IndexerThread(w, config, tw, mgr, ordLimit, stop);
indexer.Start();
try
{
while (!stop.Get())
{
SearcherAndTaxonomy pair = mgr.Acquire();
try
{
//System.out.println("search maxOrd=" + pair.taxonomyReader.getSize());
FacetsCollector sfc = new FacetsCollector();
pair.searcher.Search(new MatchAllDocsQuery(), sfc);
Facets facets = GetTaxonomyFacetCounts(pair.taxonomyReader, config, sfc);
FacetResult result = facets.GetTopChildren(10, "field");
if (pair.searcher.IndexReader.NumDocs > 0)
{
//System.out.println(pair.taxonomyReader.getSize());
Assert.True(result.ChildCount > 0);
Assert.True(result.LabelValues.Length > 0);
}
//if (VERBOSE) {
//System.out.println("TEST: facets=" + FacetTestUtils.toString(results.get(0)));
//}
}
finally
{
mgr.Release(pair);
}
}
}
finally
{
indexer.Join();
}
if (VERBOSE)
{
Console.WriteLine("TEST: now stop");
}
IOUtils.Close(mgr, tw, w, taxoDir, indexDir);
}
public virtual void TestConcurrency()
{
AtomicInteger numDocs = new AtomicInteger(AtLeast(10000));
Directory indexDir = NewDirectory();
Directory taxoDir = NewDirectory();
ConcurrentDictionary<string, string> values = new ConcurrentDictionary<string, string>();
IndexWriter iw = new IndexWriter(indexDir, NewIndexWriterConfig(TEST_VERSION_CURRENT, null));
var tw = new DirectoryTaxonomyWriter(taxoDir, OpenMode.CREATE, NewTaxoWriterCache(numDocs.Get()));
ThreadClass[] indexThreads = new ThreadClass[AtLeast(4)];
FacetsConfig config = new FacetsConfig();
for (int i = 0; i < 10; i++)
{
config.SetHierarchical("l1." + i, true);
config.SetMultiValued("l1." + i, true);
}
for (int i = 0; i < indexThreads.Length; i++)
{
indexThreads[i] = new ThreadAnonymousInnerClassHelper(this, numDocs, values, iw, tw, config);
}
foreach (ThreadClass t in indexThreads)
{
t.Start();
}
foreach (ThreadClass t in indexThreads)
{
t.Join();
}
var tr = new DirectoryTaxonomyReader(tw);
// +1 for root category
if (values.Count + 1 != tr.Count)
{
foreach (string value in values.Keys)
{
FacetLabel label = new FacetLabel(FacetsConfig.StringToPath(value));
if (tr.GetOrdinal(label) == -1)
{
Console.WriteLine("FAIL: path=" + label + " not recognized");
}
}
Fail("mismatch number of categories");
}
int[] parents = tr.ParallelTaxonomyArrays.Parents;
foreach (string cat in values.Keys)
{
FacetLabel cp = new FacetLabel(FacetsConfig.StringToPath(cat));
Assert.True(tr.GetOrdinal(cp) > 0, "category not found " + cp);
int level = cp.Length;
int parentOrd = 0; // for root, parent is always virtual ROOT (ord=0)
FacetLabel path = null;
for (int i = 0; i < level; i++)
{
path = cp.Subpath(i + 1);
int ord = tr.GetOrdinal(path);
Assert.AreEqual(parentOrd, parents[ord], "invalid parent for cp=" + path);
parentOrd = ord; // next level should have this parent
}
}
IOUtils.Close(tw, iw, tr, taxoDir, indexDir);
}
public virtual void TestLabelWithDelimiter()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir);
var taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode_e.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("dim", true);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
doc.Add(new FacetField("dim", "test\u001Fone"));
doc.Add(new FacetField("dim", "test\u001Etwo"));
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.Reader);
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
FacetsCollector c = new FacetsCollector();
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
Assert.AreEqual(1, facets.GetSpecificValue("dim", "test\u001Fone"));
Assert.AreEqual(1, facets.GetSpecificValue("dim", "test\u001Etwo"));
FacetResult result = facets.GetTopChildren(10, "dim");
Assert.AreEqual("dim=dim path=[] value=-1 childCount=2\n test\u001Fone (1)\n test\u001Etwo (1)\n", result.ToString());
IOUtils.Close(writer, taxoWriter, searcher.IndexReader, taxoReader, dir, taxoDir);
}
public virtual void TestNrt()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
IndexWriterConfig iwc = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
// Don't allow tiny maxBufferedDocs; it can make this
// test too slow:
iwc.SetMaxBufferedDocs(Math.Max(500, iwc.MaxBufferedDocs));
// MockRandom/AlcololicMergePolicy are too slow:
TieredMergePolicy tmp = new TieredMergePolicy();
tmp.FloorSegmentMB = .001;
iwc.SetMergePolicy(tmp);
IndexWriter w = new IndexWriter(dir, iwc);
var tw = new DirectoryTaxonomyWriter(taxoDir);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("field", true);
AtomicBoolean stop = new AtomicBoolean();
// How many unique facets to index before stopping:
int ordLimit = TEST_NIGHTLY ? 100000 : 6000;
var indexer = new IndexerThread(w, config, tw, null, ordLimit, stop);
var mgr = new SearcherTaxonomyManager(w, true, null, tw);
var reopener = new ThreadAnonymousInnerClassHelper(this, stop, mgr);
reopener.Name = "reopener";
reopener.Start();
indexer.Name = "indexer";
indexer.Start();
try
{
while (!stop.Get())
{
SearcherAndTaxonomy pair = mgr.Acquire();
try
{
//System.out.println("search maxOrd=" + pair.taxonomyReader.getSize());
FacetsCollector sfc = new FacetsCollector();
pair.searcher.Search(new MatchAllDocsQuery(), sfc);
Facets facets = GetTaxonomyFacetCounts(pair.taxonomyReader, config, sfc);
FacetResult result = facets.GetTopChildren(10, "field");
if (pair.searcher.IndexReader.NumDocs > 0)
{
//System.out.println(pair.taxonomyReader.getSize());
Assert.True(result.ChildCount > 0);
Assert.True(result.LabelValues.Length > 0);
}
//if (VERBOSE) {
//System.out.println("TEST: facets=" + FacetTestUtils.toString(results.get(0)));
//}
}
finally
{
mgr.Release(pair);
}
}
}
finally
{
indexer.Join();
reopener.Join();
}
if (VERBOSE)
{
Console.WriteLine("TEST: now stop");
}
IOUtils.Close(mgr, tw, w, taxoDir, dir);
}
public virtual void TestRequireDimCount()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir);
DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode_e.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetRequireDimCount("dim", true);
config.SetMultiValued("dim2", true);
config.SetRequireDimCount("dim2", true);
config.SetMultiValued("dim3", true);
config.SetHierarchical("dim3", true);
config.SetRequireDimCount("dim3", true);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
doc.Add(new FacetField("dim", "a"));
doc.Add(new FacetField("dim2", "a"));
doc.Add(new FacetField("dim2", "b"));
doc.Add(new FacetField("dim3", "a", "b"));
doc.Add(new FacetField("dim3", "a", "c"));
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.Reader);
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
FacetsCollector c = new FacetsCollector();
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim").Value);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim2").Value);
Assert.AreEqual(1, facets.GetTopChildren(10, "dim3").Value);
try
{
Assert.AreEqual(1, facets.GetSpecificValue("dim"));
Fail("didn't hit expected exception");
}
catch (System.ArgumentException)
{
// expected
}
Assert.AreEqual(1, facets.GetSpecificValue("dim2"));
Assert.AreEqual(1, facets.GetSpecificValue("dim3"));
IOUtils.Close(writer, taxoWriter, searcher.IndexReader, taxoReader, dir, taxoDir);
}
public virtual void TestDirectory()
{
Store.Directory indexDir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
IndexWriter w = new IndexWriter(indexDir, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())));
var tw = new DirectoryTaxonomyWriter(taxoDir);
// first empty commit
w.Commit();
tw.Commit();
var mgr = new SearcherTaxonomyManager(indexDir, taxoDir, null);
FacetsConfig config = new FacetsConfig();
config.SetMultiValued("field", true);
AtomicBoolean stop = new AtomicBoolean();
// How many unique facets to index before stopping:
int ordLimit = TEST_NIGHTLY ? 100000 : 6000;
var indexer = new IndexerThread(w, config, tw, mgr, ordLimit, stop);
indexer.Start();
try
{
while (!stop.Get())
{
SearcherAndTaxonomy pair = mgr.Acquire();
try
{
//System.out.println("search maxOrd=" + pair.taxonomyReader.getSize());
FacetsCollector sfc = new FacetsCollector();
pair.searcher.Search(new MatchAllDocsQuery(), sfc);
Facets facets = GetTaxonomyFacetCounts(pair.taxonomyReader, config, sfc);
FacetResult result = facets.GetTopChildren(10, "field");
if (pair.searcher.IndexReader.NumDocs > 0)
{
//System.out.println(pair.taxonomyReader.getSize());
Assert.True(result.ChildCount > 0);
Assert.True(result.LabelValues.Length > 0);
}
//if (VERBOSE) {
//System.out.println("TEST: facets=" + FacetTestUtils.toString(results.get(0)));
//}
}
finally
{
mgr.Release(pair);
}
}
}
finally
{
indexer.Join();
}
if (VERBOSE)
{
Console.WriteLine("TEST: now stop");
}
IOUtils.Close(mgr, tw, w, taxoDir, indexDir);
}
public virtual void TestMultiValuedHierarchy()
{
Store.Directory dir = NewDirectory();
Store.Directory taxoDir = NewDirectory();
DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode_e.CREATE);
FacetsConfig config = new FacetsConfig();
config.SetHierarchical("a", true);
config.SetMultiValued("a", true);
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir);
Document doc = new Document();
doc.Add(NewTextField("field", "text", Field.Store.NO));
doc.Add(new FacetField("a", "path", "x"));
doc.Add(new FacetField("a", "path", "y"));
writer.AddDocument(config.Build(taxoWriter, doc));
// NRT open
IndexSearcher searcher = NewSearcher(writer.Reader);
// NRT open
var taxoReader = new DirectoryTaxonomyReader(taxoWriter);
// Aggregate the facet counts:
FacetsCollector c = new FacetsCollector();
// MatchAllDocsQuery is for "browsing" (counts facets
// for all non-deleted docs in the index); normally
// you'd use a "normal" query, and use MultiCollector to
// wrap collecting the "normal" hits and also facets:
searcher.Search(new MatchAllDocsQuery(), c);
Facets facets = GetTaxonomyFacetCounts(taxoReader, config, c);
try
{
facets.GetSpecificValue("a");
Fail("didn't hit expected exception");
}
catch (System.ArgumentException)
{
// expected
}
FacetResult result = facets.GetTopChildren(10, "a");
Assert.AreEqual(1, result.LabelValues.Length);
Assert.AreEqual(1, (int)result.LabelValues[0].value);
IOUtils.Close(writer, taxoWriter, searcher.IndexReader, taxoReader, dir, taxoDir);
}